doc/lemon.html

   1 <html>
   2 <head>
   3 <title>The Lemon Parser Generator</title>
   4 </head>
   5 <body bgcolor=white>
   6 <h1 align=center>The Lemon Parser Generator</h1>
   7
   8 <p>Lemon is an LALR(1) parser generator for C or C++.
   9 It does the same job as ``bison'' and ``yacc''.
  10 But lemon is not another bison or yacc clone.  It
  11 uses a different grammar syntax which is designed to
  12 reduce the number of coding errors.  Lemon also uses a more
  13 sophisticated parsing engine that is faster than yacc and
  14 bison and which is both reentrant and thread-safe.
  15 Furthermore, Lemon implements features that can be used
  16 to eliminate resource leaks, making is suitable for use
  17 in long-running programs such as graphical user interfaces
  18 or embedded controllers.</p>
  19
  20 <p>This document is an introduction to the Lemon
  21 parser generator.</p>
  22
  23 <h2>Theory of Operation</h2>
  24
  25 <p>The main goal of Lemon is to translate a context free grammar (CFG)
  26 for a particular language into C code that implements a parser for
  27 that language.
  28 The program has two inputs:
  29 <ul>
  30 <li>The grammar specification.
  31 <li>A parser template file.
  32 </ul>
  33 Typically, only the grammar specification is supplied by the programmer.
  34 Lemon comes with a default parser template which works fine for most
  35 applications.  But the user is free to substitute a different parser
  36 template if desired.</p>
  37
  38 <p>Depending on command-line options, Lemon will generate between
  39 one and three files of outputs.
  40 <ul>
  41 <li>C code to implement the parser.
  42 <li>A header file defining an integer ID for each terminal symbol.
  43 <li>An information file that describes the states of the generated parser
  44     automaton.
  45 </ul>
  46 By default, all three of these output files are generated.
  47 The header file is suppressed if the ``-m'' command-line option is
  48 used and the report file is omitted when ``-q'' is selected.</p>
  49
  50 <p>The grammar specification file uses a ``.y'' suffix, by convention.
  51 In the examples used in this document, we'll assume the name of the
  52 grammar file is ``gram.y''.  A typical use of Lemon would be the
  53 following command:
  54 <pre>
  55    lemon gram.y
  56 </pre>
  57 This command will generate three output files named ``gram.c'',
  58 ``gram.h'' and ``gram.out''.
  59 The first is C code to implement the parser.  The second
  60 is the header file that defines numerical values for all
  61 terminal symbols, and the last is the report that explains
  62 the states used by the parser automaton.</p>
  63
  64 <h3>Command Line Options</h3>
  65
  66 <p>The behavior of Lemon can be modified using command-line options.
  67 You can obtain a list of the available command-line options together
  68 with a brief explanation of what each does by typing
  69 <pre>
  70    lemon -?
  71 </pre>
  72 As of this writing, the following command-line options are supported:
  73 <ul>
  74 <li><tt>-b</tt>
  75 <li><tt>-c</tt>
  76 <li><tt>-g</tt>
  77 <li><tt>-m</tt>
  78 <li><tt>-q</tt>
  79 <li><tt>-s</tt>
  80 <li><tt>-x</tt>
  81 </ul>
  82 The ``-b'' option reduces the amount of text in the report file by
  83 printing only the basis of each parser state, rather than the full
  84 configuration.
  85 The ``-c'' option suppresses action table compression.  Using -c
  86 will make the parser a little larger and slower but it will detect
  87 syntax errors sooner.
  88 The ``-g'' option causes no output files to be generated at all.
  89 Instead, the input grammar file is printed on standard output but
  90 with all comments, actions and other extraneous text deleted.  This
  91 is a useful way to get a quick summary of a grammar.
  92 The ``-m'' option causes the output C source file to be compatible
  93 with the ``makeheaders'' program.
  94 Makeheaders is a program that automatically generates header files
  95 from C source code.  When the ``-m'' option is used, the header
  96 file is not output since the makeheaders program will take care
  97 of generated all header files automatically.
  98 The ``-q'' option suppresses the report file.
  99 Using ``-s'' causes a brief summary of parser statistics to be
 100 printed.  Like this:
 101 <pre>
 102    Parser statistics: 74 terminals, 70 nonterminals, 179 rules
 103                       340 states, 2026 parser table entries, 0 conflicts
 104 </pre>
 105 Finally, the ``-x'' option causes Lemon to print its version number
 106 and then stops without attempting to read the grammar or generate a parser.</p>
 107
 108 <h3>The Parser Interface</h3>
 109
 110 <p>Lemon doesn't generate a complete, working program.  It only generates
 111 a few subroutines that implement a parser.  This section describes
 112 the interface to those subroutines.  It is up to the programmer to
 113 call these subroutines in an appropriate way in order to produce a
 114 complete system.</p>
 115
 116 <p>Before a program begins using a Lemon-generated parser, the program
 117 must first create the parser.
 118 A new parser is created as follows:
 119 <pre>
 120    void *pParser = ParseAlloc( malloc );
 121 </pre>
 122 The ParseAlloc() routine allocates and initializes a new parser and
 123 returns a pointer to it.
 124 The actual data structure used to represent a parser is opaque --
 125 its internal structure is not visible or usable by the calling routine.
 126 For this reason, the ParseAlloc() routine returns a pointer to void
 127 rather than a pointer to some particular structure.
 128 The sole argument to the ParseAlloc() routine is a pointer to the
 129 subroutine used to allocate memory.  Typically this means ``malloc()''.</p>
 130
 131 <p>After a program is finished using a parser, it can reclaim all
 132 memory allocated by that parser by calling
 133 <pre>
 134    ParseFree(pParser, free);
 135 </pre>
 136 The first argument is the same pointer returned by ParseAlloc().  The
 137 second argument is a pointer to the function used to release bulk
 138 memory back to the system.</p>
 139
 140 <p>After a parser has been allocated using ParseAlloc(), the programmer
 141 must supply the parser with a sequence of tokens (terminal symbols) to
 142 be parsed.  This is accomplished by calling the following function
 143 once for each token:
 144 <pre>
 145    Parse(pParser, hTokenID, sTokenData, pArg);
 146 </pre>
 147 The first argument to the Parse() routine is the pointer returned by
 148 ParseAlloc().
 149 The second argument is a small positive integer that tells the parse the
 150 type of the next token in the data stream.
 151 There is one token type for each terminal symbol in the grammar.
 152 The gram.h file generated by Lemon contains #define statements that
 153 map symbolic terminal symbol names into appropriate integer values.
 154 (A value of 0 for the second argument is a special flag to the
 155 parser to indicate that the end of input has been reached.)
 156 The third argument is the value of the given token.  By default,
 157 the type of the third argument is integer, but the grammar will
 158 usually redefine this type to be some kind of structure.
 159 Typically the second argument will be a broad category of tokens
 160 such as ``identifier'' or ``number'' and the third argument will
 161 be the name of the identifier or the value of the number.</p>
 162
 163 <p>The Parse() function may have either three or four arguments,
 164 depending on the grammar.  If the grammar specification file request
 165 it, the Parse() function will have a fourth parameter that can be
 166 of any type chosen by the programmer.  The parser doesn't do anything
 167 with this argument except to pass it through to action routines.
 168 This is a convenient mechanism for passing state information down
 169 to the action routines without having to use global variables.</p>
 170
 171 <p>A typical use of a Lemon parser might look something like the
 172 following:
 173 <pre>
 174    01 ParseTree *ParseFile(const char *zFilename){
 175    02    Tokenizer *pTokenizer;
 176    03    void *pParser;
 177    04    Token sToken;
 178    05    int hTokenId;
 179    06    ParserState sState;
 180    07
 181    08    pTokenizer = TokenizerCreate(zFilename);
 182    09    pParser = ParseAlloc( malloc );
 183    10    InitParserState(&sState);
 184    11    while( GetNextToken(pTokenizer, &hTokenId, &sToken) ){
 185    12       Parse(pParser, hTokenId, sToken, &sState);
 186    13    }
 187    14    Parse(pParser, 0, sToken, &sState);
 188    15    ParseFree(pParser, free );
 189    16    TokenizerFree(pTokenizer);
 190    17    return sState.treeRoot;
 191    18 }
 192 </pre>
 193 This example shows a user-written routine that parses a file of
 194 text and returns a pointer to the parse tree.
 195 (We've omitted all error-handling from this example to keep it
 196 simple.)
 197 We assume the existence of some kind of tokenizer which is created
 198 using TokenizerCreate() on line 8 and deleted by TokenizerFree()
 199 on line 16.  The GetNextToken() function on line 11 retrieves the
 200 next token from the input file and puts its type in the
 201 integer variable hTokenId.  The sToken variable is assumed to be
 202 some kind of structure that contains details about each token,
 203 such as its complete text, what line it occurs on, etc. </p>
 204
 205 <p>This example also assumes the existence of structure of type
 206 ParserState that holds state information about a particular parse.
 207 An instance of such a structure is created on line 6 and initialized
 208 on line 10.  A pointer to this structure is passed into the Parse()
 209 routine as the optional 4th argument.
 210 The action routine specified by the grammar for the parser can use
 211 the ParserState structure to hold whatever information is useful and
 212 appropriate.  In the example, we note that the treeRoot field of
 213 the ParserState structure is left pointing to the root of the parse
 214 tree.</p>
 215
 216 <p>The core of this example as it relates to Lemon is as follows:
 217 <pre>
 218    ParseFile(){
 219       pParser = ParseAlloc( malloc );
 220       while( GetNextToken(pTokenizer,&hTokenId, &sToken) ){
 221          Parse(pParser, hTokenId, sToken);
 222       }
 223       Parse(pParser, 0, sToken);
 224       ParseFree(pParser, free );
 225    }
 226 </pre>
 227 Basically, what a program has to do to use a Lemon-generated parser
 228 is first create the parser, then send it lots of tokens obtained by
 229 tokenizing an input source.  When the end of input is reached, the
 230 Parse() routine should be called one last time with a token type
 231 of 0.  This step is necessary to inform the parser that the end of
 232 input has been reached.  Finally, we reclaim memory used by the
 233 parser by calling ParseFree().</p>
 234
 235 <p>There is one other interface routine that should be mentioned
 236 before we move on.
 237 The ParseTrace() function can be used to generate debugging output
 238 from the parser.  A prototype for this routine is as follows:
 239 <pre>
 240    ParseTrace(FILE *stream, char *zPrefix);
 241 </pre>
 242 After this routine is called, a short (one-line) message is written
 243 to the designated output stream every time the parser changes states
 244 or calls an action routine.  Each such message is prefaced using
 245 the text given by zPrefix.  This debugging output can be turned off
 246 by calling ParseTrace() again with a first argument of NULL (0).</p>
 247
 248 <h3>Differences With YACC and BISON</h3>
 249
 250 <p>Programmers who have previously used the yacc or bison parser
 251 generator will notice several important differences between yacc and/or
 252 bison and Lemon.
 253 <ul>
 254 <li>In yacc and bison, the parser calls the tokenizer.  In Lemon,
 255     the tokenizer calls the parser.
 256 <li>Lemon uses no global variables.  Yacc and bison use global variables
 257     to pass information between the tokenizer and parser.
 258 <li>Lemon allows multiple parsers to be running simultaneously.  Yacc
 259     and bison do not.
 260 </ul>
 261 These differences may cause some initial confusion for programmers
 262 with prior yacc and bison experience.
 263 But after years of experience using Lemon, I firmly
 264 believe that the Lemon way of doing things is better.</p>
 265
 266 <h2>Input File Syntax</h2>
 267
 268 <p>The main purpose of the grammar specification file for Lemon is
 269 to define the grammar for the parser.  But the input file also
 270 specifies additional information Lemon requires to do its job.
 271 Most of the work in using Lemon is in writing an appropriate
 272 grammar file.</p>
 273
 274 <p>The grammar file for lemon is, for the most part, free format.
 275 It does not have sections or divisions like yacc or bison.  Any
 276 declaration can occur at any point in the file.
 277 Lemon ignores whitespace (except where it is needed to separate
 278 tokens) and it honors the same commenting conventions as C and C++.</p>
 279
 280 <h3>Terminals and Nonterminals</h3>
 281
 282 <p>A terminal symbol (token) is any string of alphanumeric
 283 and underscore characters
 284 that begins with an upper case letter.
 285 A terminal can contain lowercase letters after the first character,
 286 but the usual convention is to make terminals all upper case.
 287 A nonterminal, on the other hand, is any string of alphanumeric
 288 and underscore characters than begins with a lower case letter.
 289 Again, the usual convention is to make nonterminals use all lower
 290 case letters.</p>
 291
 292 <p>In Lemon, terminal and nonterminal symbols do not need to
 293 be declared or identified in a separate section of the grammar file.
 294 Lemon is able to generate a list of all terminals and nonterminals
 295 by examining the grammar rules, and it can always distinguish a
 296 terminal from a nonterminal by checking the case of the first
 297 character of the name.</p>
 298
 299 <p>Yacc and bison allow terminal symbols to have either alphanumeric
 300 names or to be individual characters included in single quotes, like
 301 this: ')' or '$'.  Lemon does not allow this alternative form for
 302 terminal symbols.  With Lemon, all symbols, terminals and nonterminals,
 303 must have alphanumeric names.</p>
 304
 305 <h3>Grammar Rules</h3>
 306
 307 <p>The main component of a Lemon grammar file is a sequence of grammar
 308 rules.
 309 Each grammar rule consists of a nonterminal symbol followed by
 310 the special symbol ``::='' and then a list of terminals and/or nonterminals.
 311 The rule is terminated by a period.
 312 The list of terminals and nonterminals on the right-hand side of the
 313 rule can be empty.
 314 Rules can occur in any order, except that the left-hand side of the
 315 first rule is assumed to be the start symbol for the grammar (unless
 316 specified otherwise using the <tt>%start</tt> directive described below.)
 317 A typical sequence of grammar rules might look something like this:
 318 <pre>
 319   expr ::= expr PLUS expr.
 320   expr ::= expr TIMES expr.
 321   expr ::= LPAREN expr RPAREN.
 322   expr ::= VALUE.
 323 </pre>
 324 </p>
 325
 326 <p>There is one non-terminal in this example, ``expr'', and five
 327 terminal symbols or tokens: ``PLUS'', ``TIMES'', ``LPAREN'',
 328 ``RPAREN'' and ``VALUE''.</p>
 329
 330 <p>Like yacc and bison, Lemon allows the grammar to specify a block
 331 of C code that will be executed whenever a grammar rule is reduced
 332 by the parser.
 333 In Lemon, this action is specified by putting the C code (contained
 334 within curly braces <tt>{...}</tt>) immediately after the
 335 period that closes the rule.
 336 For example:
 337 <pre>
 338   expr ::= expr PLUS expr.   { printf("Doing an addition...\n"); }
 339 </pre>
 340 </p>
 341
 342 <p>In order to be useful, grammar actions must normally be linked to
 343 their associated grammar rules.
 344 In yacc and bison, this is accomplished by embedding a ``$$'' in the
 345 action to stand for the value of the left-hand side of the rule and
 346 symbols ``$1'', ``$2'', and so forth to stand for the value of
 347 the terminal or nonterminal at position 1, 2 and so forth on the
 348 right-hand side of the rule.
 349 This idea is very powerful, but it is also very error-prone.  The
 350 single most common source of errors in a yacc or bison grammar is
 351 to miscount the number of symbols on the right-hand side of a grammar
 352 rule and say ``$7'' when you really mean ``$8''.</p>
 353
 354 <p>Lemon avoids the need to count grammar symbols by assigning symbolic
 355 names to each symbol in a grammar rule and then using those symbolic
 356 names in the action.
 357 In yacc or bison, one would write this:
 358 <pre>
 359   expr -> expr PLUS expr  { $$ = $1 + $3; };
 360 </pre>
 361 But in Lemon, the same rule becomes the following:
 362 <pre>
 363   expr(A) ::= expr(B) PLUS expr(C).  { A = B+C; }
 364 </pre>
 365 In the Lemon rule, any symbol in parentheses after a grammar rule
 366 symbol becomes a place holder for that symbol in the grammar rule.
 367 This place holder can then be used in the associated C action to
 368 stand for the value of that symbol.<p>
 369
 370 <p>The Lemon notation for linking a grammar rule with its reduce
 371 action is superior to yacc/bison on several counts.
 372 First, as mentioned above, the Lemon method avoids the need to
 373 count grammar symbols.
 374 Secondly, if a terminal or nonterminal in a Lemon grammar rule
 375 includes a linking symbol in parentheses but that linking symbol
 376 is not actually used in the reduce action, then an error message
 377 is generated.
 378 For example, the rule
 379 <pre>
 380   expr(A) ::= expr(B) PLUS expr(C).  { A = B; }
 381 </pre>
 382 will generate an error because the linking symbol ``C'' is used
 383 in the grammar rule but not in the reduce action.</p>
 384
 385 <p>The Lemon notation for linking grammar rules to reduce actions
 386 also facilitates the use of destructors for reclaiming memory
 387 allocated by the values of terminals and nonterminals on the
 388 right-hand side of a rule.</p>
 389
 390 <h3>Precedence Rules</h3>
 391
 392 <p>Lemon resolves parsing ambiguities in exactly the same way as
 393 yacc and bison.  A shift-reduce conflict is resolved in favor
 394 of the shift, and a reduce-reduce conflict is resolved by reducing
 395 whichever rule comes first in the grammar file.</p>
 396
 397 <p>Just like in
 398 yacc and bison, Lemon allows a measure of control
 399 over the resolution of paring conflicts using precedence rules.
 400 A precedence value can be assigned to any terminal symbol
 401 using the %left, %right or %nonassoc directives.  Terminal symbols
 402 mentioned in earlier directives have a lower precedence that
 403 terminal symbols mentioned in later directives.  For example:</p>
 404
 405 <p><pre>
 406    %left AND.
 407    %left OR.
 408    %nonassoc EQ NE GT GE LT LE.
 409    %left PLUS MINUS.
 410    %left TIMES DIVIDE MOD.
 411    %right EXP NOT.
 412 </pre></p>
 413
 414 <p>In the preceding sequence of directives, the AND operator is
 415 defined to have the lowest precedence.  The OR operator is one
 416 precedence level higher.  And so forth.  Hence, the grammar would
 417 attempt to group the ambiguous expression
 418 <pre>
 419      a AND b OR c
 420 </pre>
 421 like this
 422 <pre>
 423      a AND (b OR c).
 424 </pre>
 425 The associativity (left, right or nonassoc) is used to determine
 426 the grouping when the precedence is the same.  AND is left-associative
 427 in our example, so
 428 <pre>
 429      a AND b AND c
 430 </pre>
 431 is parsed like this
 432 <pre>
 433      (a AND b) AND c.
 434 </pre>
 435 The EXP operator is right-associative, though, so
 436 <pre>
 437      a EXP b EXP c
 438 </pre>
 439 is parsed like this
 440 <pre>
 441      a EXP (b EXP c).
 442 </pre>
 443 The nonassoc precedence is used for non-associative operators.
 444 So
 445 <pre>
 446      a EQ b EQ c
 447 </pre>
 448 is an error.</p>
 449
 450 <p>The precedence of non-terminals is transferred to rules as follows:
 451 The precedence of a grammar rule is equal to the precedence of the
 452 left-most terminal symbol in the rule for which a precedence is
 453 defined.  This is normally what you want, but in those cases where
 454 you want to precedence of a grammar rule to be something different,
 455 you can specify an alternative precedence symbol by putting the
 456 symbol in square braces after the period at the end of the rule and
 457 before any C-code.  For example:</p>
 458
 459 <p><pre>
 460    expr = MINUS expr.  [NOT]
 461 </pre></p>
 462
 463 <p>This rule has a precedence equal to that of the NOT symbol, not the
 464 MINUS symbol as would have been the case by default.</p>
 465
 466 <p>With the knowledge of how precedence is assigned to terminal
 467 symbols and individual
 468 grammar rules, we can now explain precisely how parsing conflicts
 469 are resolved in Lemon.  Shift-reduce conflicts are resolved
 470 as follows:
 471 <ul>
 472 <li> If either the token to be shifted or the rule to be reduced
 473      lacks precedence information, then resolve in favor of the
 474      shift, but report a parsing conflict.
 475 <li> If the precedence of the token to be shifted is greater than
 476      the precedence of the rule to reduce, then resolve in favor
 477      of the shift.  No parsing conflict is reported.
 478 <li> If the precedence of the token it be shifted is less than the
 479      precedence of the rule to reduce, then resolve in favor of the
 480      reduce action.  No parsing conflict is reported.
 481 <li> If the precedences are the same and the shift token is
 482      right-associative, then resolve in favor of the shift.
 483      No parsing conflict is reported.
 484 <li> If the precedences are the same the shift token is
 485      left-associative, then resolve in favor of the reduce.
 486      No parsing conflict is reported.
 487 <li> Otherwise, resolve the conflict by doing the shift and
 488      report the parsing conflict.
 489 </ul>
 490 Reduce-reduce conflicts are resolved this way:
 491 <ul>
 492 <li> If either reduce rule
 493      lacks precedence information, then resolve in favor of the
 494      rule that appears first in the grammar and report a parsing
 495      conflict.
 496 <li> If both rules have precedence and the precedence is different
 497      then resolve the dispute in favor of the rule with the highest
 498      precedence and do not report a conflict.
 499 <li> Otherwise, resolve the conflict by reducing by the rule that
 500      appears first in the grammar and report a parsing conflict.
 501 </ul>
 502
 503 <h3>Special Directives</h3>
 504
 505 <p>The input grammar to Lemon consists of grammar rules and special
 506 directives.  We've described all the grammar rules, so now we'll
 507 talk about the special directives.</p>
 508
 509 <p>Directives in lemon can occur in any order.  You can put them before
 510 the grammar rules, or after the grammar rules, or in the mist of the
 511 grammar rules.  It doesn't matter.  The relative order of
 512 directives used to assign precedence to terminals is important, but
 513 other than that, the order of directives in Lemon is arbitrary.</p>
 514
 515 <p>Lemon supports the following special directives:
 516 <ul>
 517 <li><tt>%code</tt>
 518 <li><tt>%default_destructor</tt>
 519 <li><tt>%default_type</tt>
 520 <li><tt>%destructor</tt>
 521 <li><tt>%extra_argument</tt>
 522 <li><tt>%include</tt>
 523 <li><tt>%left</tt>
 524 <li><tt>%name</tt>
 525 <li><tt>%nonassoc</tt>
 526 <li><tt>%parse_accept</tt>
 527 <li><tt>%parse_failure </tt>
 528 <li><tt>%right</tt>
 529 <li><tt>%stack_overflow</tt>
 530 <li><tt>%stack_size</tt>
 531 <li><tt>%start_symbol</tt>
 532 <li><tt>%syntax_error</tt>
 533 <li><tt>%token_destructor</tt>
 534 <li><tt>%token_prefix</tt>
 535 <li><tt>%token_type</tt>
 536 <li><tt>%type</tt>
 537 </ul>
 538 Each of these directives will be described separately in the
 539 following sections:</p>
 540
 541 <h4>The <tt>%code</tt> directive</h4>
 542
 543 <p>The %code directive is used to specify addition C/C++ code that
 544 is added to the end of the main output file.  This is similar to
 545 the %include directive except that %include is inserted at the
 546 beginning of the main output file.</p>
 547
 548 <p>%code is typically used to include some action routines or perhaps
 549 a tokenizer as part of the output file.</p>
 550
 551 <h4>The <tt>%default_destructor</tt> directive</h4>
 552
 553 <p>The %default_destructor directive specifies a destructor to
 554 use for non-terminals that do not have their own destructor
 555 specified by a separate %destructor directive.  See the documentation
 556 on the %destructor directive below for additional information.</p>
 557
 558 <p>In some grammers, many different non-terminal symbols have the
 559 same datatype and hence the same destructor.  This directive is
 560 a convenience way to specify the same destructor for all those
 561 non-terminals using a single statement.</p>
 562
 563 <h4>The <tt>%default_type</tt> directive</h4>
 564
 565 <p>The %default_type directive specifies the datatype of non-terminal
 566 symbols that do no have their own datatype defined using a separate
 567 %type directive.  See the documentation on %type below for addition
 568 information.</p>
 569
 570 <h4>The <tt>%destructor</tt> directive</h4>
 571
 572 <p>The %destructor directive is used to specify a destructor for
 573 a non-terminal symbol.
 574 (See also the %token_destructor directive which is used to
 575 specify a destructor for terminal symbols.)</p>
 576
 577 <p>A non-terminal's destructor is called to dispose of the
 578 non-terminal's value whenever the non-terminal is popped from
 579 the stack.  This includes all of the following circumstances:
 580 <ul>
 581 <li> When a rule reduces and the value of a non-terminal on
 582      the right-hand side is not linked to C code.
 583 <li> When the stack is popped during error processing.
 584 <li> When the ParseFree() function runs.
 585 </ul>
 586 The destructor can do whatever it wants with the value of
 587 the non-terminal, but its design is to deallocate memory
 588 or other resources held by that non-terminal.</p>
 589
 590 <p>Consider an example:
 591 <pre>
 592    %type nt {void*}
 593    %destructor nt { free($$); }
 594    nt(A) ::= ID NUM.   { A = malloc( 100 ); }
 595 </pre>
 596 This example is a bit contrived but it serves to illustrate how
 597 destructors work.  The example shows a non-terminal named
 598 ``nt'' that holds values of type ``void*''.  When the rule for
 599 an ``nt'' reduces, it sets the value of the non-terminal to
 600 space obtained from malloc().  Later, when the nt non-terminal
 601 is popped from the stack, the destructor will fire and call
 602 free() on this malloced space, thus avoiding a memory leak.
 603 (Note that the symbol ``$$'' in the destructor code is replaced
 604 by the value of the non-terminal.)</p>
 605
 606 <p>It is important to note that the value of a non-terminal is passed
 607 to the destructor whenever the non-terminal is removed from the
 608 stack, unless the non-terminal is used in a C-code action.  If
 609 the non-terminal is used by C-code, then it is assumed that the
 610 C-code will take care of destroying it if it should really
 611 be destroyed.  More commonly, the value is used to build some
 612 larger structure and we don't want to destroy it, which is why
 613 the destructor is not called in this circumstance.</p>
 614
 615 <p>By appropriate use of destructors, it is possible to
 616 build a parser using Lemon that can be used within a long-running
 617 program, such as a GUI, that will not leak memory or other resources.
 618 To do the same using yacc or bison is much more difficult.</p>
 619
 620 <h4>The <tt>%extra_argument</tt> directive</h4>
 621
 622 The %extra_argument directive instructs Lemon to add a 4th parameter
 623 to the parameter list of the Parse() function it generates.  Lemon
 624 doesn't do anything itself with this extra argument, but it does
 625 make the argument available to C-code action routines, destructors,
 626 and so forth.  For example, if the grammar file contains:</p>
 627
 628 <p><pre>
 629     %extra_argument { MyStruct *pAbc }
 630 </pre></p>
 631
 632 <p>Then the Parse() function generated will have an 4th parameter
 633 of type ``MyStruct*'' and all action routines will have access to
 634 a variable named ``pAbc'' that is the value of the 4th parameter
 635 in the most recent call to Parse().</p>
 636
 637 <h4>The <tt>%include</tt> directive</h4>
 638
 639 <p>The %include directive specifies C code that is included at the
 640 top of the generated parser.  You can include any text you want --
 641 the Lemon parser generator copies it blindly.  If you have multiple
 642 %include directives in your grammar file the value of the last
 643 %include directive overwrites all the others.</p.
 644
 645 <p>The %include directive is very handy for getting some extra #include
 646 preprocessor statements at the beginning of the generated parser.
 647 For example:</p>
 648
 649 <p><pre>
 650    %include {#include &lt;unistd.h&gt;}
 651 </pre></p>
 652
 653 <p>This might be needed, for example, if some of the C actions in the
 654 grammar call functions that are prototyed in unistd.h.</p>
 655
 656 <h4>The <tt>%left</tt> directive</h4>
 657
 658 The %left directive is used (along with the %right and
 659 %nonassoc directives) to declare precedences of terminal
 660 symbols.  Every terminal symbol whose name appears after
 661 a %left directive but before the next period (``.'') is
 662 given the same left-associative precedence value.  Subsequent
 663 %left directives have higher precedence.  For example:</p>
 664
 665 <p><pre>
 666    %left AND.
 667    %left OR.
 668    %nonassoc EQ NE GT GE LT LE.
 669    %left PLUS MINUS.
 670    %left TIMES DIVIDE MOD.
 671    %right EXP NOT.
 672 </pre></p>
 673
 674 <p>Note the period that terminates each %left, %right or %nonassoc
 675 directive.</p>
 676
 677 <p>LALR(1) grammars can get into a situation where they require
 678 a large amount of stack space if you make heavy use or right-associative
 679 operators.  For this reason, it is recommended that you use %left
 680 rather than %right whenever possible.</p>
 681
 682 <h4>The <tt>%name</tt> directive</h4>
 683
 684 <p>By default, the functions generated by Lemon all begin with the
 685 five-character string ``Parse''.  You can change this string to something
 686 different using the %name directive.  For instance:</p>
 687
 688 <p><pre>
 689    %name Abcde
 690 </pre></p>
 691
 692 <p>Putting this directive in the grammar file will cause Lemon to generate
 693 functions named
 694 <ul>
 695 <li> AbcdeAlloc(),
 696 <li> AbcdeFree(),
 697 <li> AbcdeTrace(), and
 698 <li> Abcde().
 699 </ul>
 700 The %name directive allows you to generator two or more different
 701 parsers and link them all into the same executable.
 702 </p>
 703
 704 <h4>The <tt>%nonassoc</tt> directive</h4>
 705
 706 <p>This directive is used to assign non-associative precedence to
 707 one or more terminal symbols.  See the section on precedence rules
 708 or on the %left directive for additional information.</p>
 709
 710 <h4>The <tt>%parse_accept</tt> directive</h4>
 711
 712 <p>The %parse_accept directive specifies a block of C code that is
 713 executed whenever the parser accepts its input string.  To ``accept''
 714 an input string means that the parser was able to process all tokens
 715 without error.</p>
 716
 717 <p>For example:</p>
 718
 719 <p><pre>
 720    %parse_accept {
 721       printf("parsing complete!\n");
 722    }
 723 </pre></p>
 724
 725
 726 <h4>The <tt>%parse_failure</tt> directive</h4>
 727
 728 <p>The %parse_failure directive specifies a block of C code that
 729 is executed whenever the parser fails complete.  This code is not
 730 executed until the parser has tried and failed to resolve an input
 731 error using is usual error recovery strategy.  The routine is
 732 only invoked when parsing is unable to continue.</p>
 733
 734 <p><pre>
 735    %parse_failure {
 736      fprintf(stderr,"Giving up.  Parser is hopelessly lost...\n");
 737    }
 738 </pre></p>
 739
 740 <h4>The <tt>%right</tt> directive</h4>
 741
 742 <p>This directive is used to assign right-associative precedence to
 743 one or more terminal symbols.  See the section on precedence rules
 744 or on the %left directive for additional information.</p>
 745
 746 <h4>The <tt>%stack_overflow</tt> directive</h4>
 747
 748 <p>The %stack_overflow directive specifies a block of C code that
 749 is executed if the parser's internal stack ever overflows.  Typically
 750 this just prints an error message.  After a stack overflow, the parser
 751 will be unable to continue and must be reset.</p>
 752
 753 <p><pre>
 754    %stack_overflow {
 755      fprintf(stderr,"Giving up.  Parser stack overflow\n");
 756    }
 757 </pre></p>
 758
 759 <p>You can help prevent parser stack overflows by avoiding the use
 760 of right recursion and right-precedence operators in your grammar.
 761 Use left recursion and and left-precedence operators instead, to
 762 encourage rules to reduce sooner and keep the stack size down.
 763 For example, do rules like this:
 764 <pre>
 765    list ::= list element.      // left-recursion.  Good!
 766    list ::= .
 767 </pre>
 768 Not like this:
 769 <pre>
 770    list ::= element list.      // right-recursion.  Bad!
 771    list ::= .
 772 </pre>
 773
 774 <h4>The <tt>%stack_size</tt> directive</h4>
 775
 776 <p>If stack overflow is a problem and you can't resolve the trouble
 777 by using left-recursion, then you might want to increase the size
 778 of the parser's stack using this directive.  Put an positive integer
 779 after the %stack_size directive and Lemon will generate a parse
 780 with a stack of the requested size.  The default value is 100.</p>
 781
 782 <p><pre>
 783    %stack_size 2000
 784 </pre></p>
 785
 786 <h4>The <tt>%start_symbol</tt> directive</h4>
 787
 788 <p>By default, the start-symbol for the grammar that Lemon generates
 789 is the first non-terminal that appears in the grammar file.  But you
 790 can choose a different start-symbol using the %start_symbol directive.</p>
 791
 792 <p><pre>
 793    %start_symbol  prog
 794 </pre></p>
 795
 796 <h4>The <tt>%token_destructor</tt> directive</h4>
 797
 798 <p>The %destructor directive assigns a destructor to a non-terminal
 799 symbol.  (See the description of the %destructor directive above.)
 800 This directive does the same thing for all terminal symbols.</p>
 801
 802 <p>Unlike non-terminal symbols which may each have a different data type
 803 for their values, terminals all use the same data type (defined by
 804 the %token_type directive) and so they use a common destructor.  Other
 805 than that, the token destructor works just like the non-terminal
 806 destructors.</p>
 807
 808 <h4>The <tt>%token_prefix</tt> directive</h4>
 809
 810 <p>Lemon generates #defines that assign small integer constants
 811 to each terminal symbol in the grammar.  If desired, Lemon will
 812 add a prefix specified by this directive
 813 to each of the #defines it generates.
 814 So if the default output of Lemon looked like this:
 815 <pre>
 816     #define AND              1
 817     #define MINUS            2
 818     #define OR               3
 819     #define PLUS             4
 820 </pre>
 821 You can insert a statement into the grammar like this:
 822 <pre>
 823     %token_prefix    TOKEN_
 824 </pre>
 825 to cause Lemon to produce these symbols instead:
 826 <pre>
 827     #define TOKEN_AND        1
 828     #define TOKEN_MINUS      2
 829     #define TOKEN_OR         3
 830     #define TOKEN_PLUS       4
 831 </pre>
 832
 833 <h4>The <tt>%token_type</tt> and <tt>%type</tt> directives</h4>
 834
 835 <p>These directives are used to specify the data types for values
 836 on the parser's stack associated with terminal and non-terminal
 837 symbols.  The values of all terminal symbols must be of the same
 838 type.  This turns out to be the same data type as the 3rd parameter
 839 to the Parse() function generated by Lemon.  Typically, you will
 840 make the value of a terminal symbol by a pointer to some kind of
 841 token structure.  Like this:</p>
 842
 843 <p><pre>
 844    %token_type    {Token*}
 845 </pre></p>
 846
 847 <p>If the data type of terminals is not specified, the default value
 848 is ``int''.</p>
 849
 850 <p>Non-terminal symbols can each have their own data types.  Typically
 851 the data type  of a non-terminal is a pointer to the root of a parse-tree
 852 structure that contains all information about that non-terminal.
 853 For example:</p>
 854
 855 <p><pre>
 856    %type   expr  {Expr*}
 857 </pre></p>
 858
 859 <p>Each entry on the parser's stack is actually a union containing
 860 instances of all data types for every non-terminal and terminal symbol.
 861 Lemon will automatically use the correct element of this union depending
 862 on what the corresponding non-terminal or terminal symbol is.  But
 863 the grammar designer should keep in mind that the size of the union
 864 will be the size of its largest element.  So if you have a single
 865 non-terminal whose data type requires 1K of storage, then your 100
 866 entry parser stack will require 100K of heap space.  If you are willing
 867 and able to pay that price, fine.  You just need to know.</p>
 868
 869 <h3>Error Processing</h3>
 870
 871 <p>After extensive experimentation over several years, it has been
 872 discovered that the error recovery strategy used by yacc is about
 873 as good as it gets.  And so that is what Lemon uses.</p>
 874
 875 <p>When a Lemon-generated parser encounters a syntax error, it
 876 first invokes the code specified by the %syntax_error directive, if
 877 any.  It then enters its error recovery strategy.  The error recovery
 878 strategy is to begin popping the parsers stack until it enters a
 879 state where it is permitted to shift a special non-terminal symbol
 880 named ``error''.  It then shifts this non-terminal and continues
 881 parsing.  But the %syntax_error routine will not be called again
 882 until at least three new tokens have been successfully shifted.</p>
 883
 884 <p>If the parser pops its stack until the stack is empty, and it still
 885 is unable to shift the error symbol, then the %parse_failed routine
 886 is invoked and the parser resets itself to its start state, ready
 887 to begin parsing a new file.  This is what will happen at the very
 888 first syntax error, of course, if there are no instances of the
 889 ``error'' non-terminal in your grammar.</p>
 890
 891 </body>
 892 </html>