1 \input texinfo @c -*-texinfo-*-
3 @setfilename ../../info/wisent
4 @set TITLE Wisent Parser Development
5 @set AUTHOR Eric M. Ludlam, David Ponce, and Richard Y. Kim
6 @settitle @value{TITLE}
8 @c *************************************************************************
10 @c *************************************************************************
12 @c Merge all indexes into a single index for now.
13 @c We can always separate them later into two or more as needed.
20 @c @footnotestyle separate
26 Copyright @copyright{} 1988--1993, 1995, 1998--2004, 2007, 2012--2013
27 Free Software Foundation, Inc.
29 @c Since we are both GNU manuals, we do not need to ack each other here.
31 Some texts are borrowed or adapted from the manual of Bison version
32 1.35. The text in section entitled ``Understanding the automaton'' is
33 adapted from the section ``Understanding Your Parser'' in the manual
34 of Bison version 1.49.
38 Permission is granted to copy, distribute and/or modify this document
39 under the terms of the GNU Free Documentation License, Version 1.3 or
40 any later version published by the Free Software Foundation; with no
41 Invariant Sections, with the Front-Cover texts being ``A GNU Manual,''
42 and with the Back-Cover Texts as in (a) below. A copy of the license
43 is included in the section entitled ``GNU Free Documentation License''.
45 (a) The FSF's Back-Cover Text is: ``You have the freedom to copy and
46 modify this GNU manual.''
50 @dircategory Emacs misc features
52 * Wisent: (wisent). Semantic Wisent parser development.
59 @c @setchapternewpage odd
60 @c @setchapternewpage off
65 @author by @value{AUTHOR}
67 @vskip 0pt plus 1 fill
76 @c *************************************************************************
78 @c *************************************************************************
84 Wisent (the European Bison ;-) is an Emacs Lisp implementation of the
85 GNU Compiler Compiler Bison.
87 This manual describes how to use Wisent to develop grammars for
88 programming languages, and how to use grammars to parse language
89 source in Emacs buffers.
91 It also describes how Wisent is used with the @semantic{} tool set
92 described in the @ref{Top, Semantic Manual, Semantic Manual, semantic}.
103 * GNU Free Documentation License::
107 @node Wisent Overview
108 @chapter Wisent Overview
110 @dfn{Wisent} (the European Bison) is an implementation in Emacs Lisp
111 of the GNU Compiler Compiler Bison. Its code is a port of the C code
112 of GNU Bison 1.28 & 1.31.
114 For more details on the basic concepts for understanding Wisent, it is
115 worthwhile to read the @ref{Top, Bison Manual, , bison}.
117 @uref{http://www.gnu.org/manual/bison/html_node/index.html}.
120 Wisent can generate compilers compatible with the @semantic{} tool set.
121 See the @ref{Top, Semantic Manual, , semantic}.
123 It benefits from these Bison features:
127 It uses a fast but not so space-efficient encoding for the parse
128 tables, described in Corbett's PhD thesis from Berkeley:
130 @cite{Static Semantics in Compiler Error Recovery}@*
131 June 1985, Report No. UCB/CSD 85/251.
135 For generating the lookahead sets, Wisent uses the well-known
136 technique of F. DeRemer and A. Pennello they described in:
138 @cite{Efficient Construction of LALR(1) Lookahead Sets}@*
139 October 1982, ACM TOPLS Vol 4 No 4.
143 Wisent resolves shift/reduce conflicts using operator precedence and
147 Parser error recovery is accomplished using rules which match the
148 special token @code{error}.
151 Nevertheless there are some fundamental differences between Bison and
156 Wisent is intended to be used in Emacs. It reads and produces Emacs
157 Lisp data structures. All the additional code used in grammars is
161 Contrary to Bison, Wisent does not generate a parser which combines
162 Emacs Lisp code and grammar constructs. They exist separately.
163 Wisent reads the grammar from a Lisp data structure and then generates
164 grammar constructs as tables. Afterward, the derived tables can be
165 included and byte-compiled in separate Emacs Lisp files, and be used
166 at a later time by the Wisent's parser engine.
169 Wisent allows multiple start nonterminals and allows a call to the
170 parsing function to be made for a particular start nonterminal. For
171 example, this is particularly useful to parse a region of an Emacs
172 buffer. @semantic{} heavily depends on the availability of this feature.
176 @chapter Wisent Grammar
178 @cindex context-free grammar
180 In order for Wisent to parse a language, it must be described by a
181 @dfn{context-free grammar}. That is a grammar specified as rules that
182 can be applied regardless of context. For more information, see
183 @ref{Language and Grammar, , , bison}, in the Bison manual.
187 The formal grammar is formulated using @dfn{terminal} and
188 @dfn{nonterminal} items. Terminals can be Emacs Lisp symbols or
189 characters, and nonterminals are symbols only.
192 Terminals (also known as @dfn{tokens}) represent the lexical
193 elements of the language like numbers, strings, etc..
195 For example @samp{PLUS} can represent the operator @samp{+}.
197 Nonterminal symbols are described by rules:
201 RESULT @equiv{} COMPONENTS@dots{}
205 @samp{RESULT} is a nonterminal that this rule describes and
206 @samp{COMPONENTS} are various terminals and nonterminals that are put
207 together by this rule.
209 For example, this rule:
213 exp @equiv{} exp PLUS exp
217 Says that two groupings of type @samp{exp}, with a @samp{PLUS} token
218 in between, can be combined into a larger grouping of type @samp{exp}.
223 * Compiling a grammar::
227 @node Grammar format, Example, Wisent Grammar, Wisent Grammar
228 @comment node-name, next, previous, up
229 @section Grammar format
231 @cindex grammar format
232 To be acceptable by Wisent a context-free grammar must respect a
233 particular format. That is, must be represented as an Emacs Lisp list
236 @code{(@var{terminals} @var{assocs} . @var{non-terminals})}
240 Is the list of terminal symbols used in the grammar.
242 @cindex associativity
244 Specify the associativity of @var{terminals}. It is @code{nil} when
245 there is no associativity defined, or an alist of
246 @w{@code{(@var{assoc-type} . @var{assoc-value})}} elements.
248 @var{assoc-type} must be one of the @code{default-prec},
249 @code{nonassoc}, @code{left} or @code{right} symbols. When
250 @var{assoc-type} is @code{default-prec}, @var{assoc-value} must be
251 @code{nil} or @code{t} (the default). Otherwise it is a list of
252 tokens which must have been previously declared in @var{terminals}.
254 For details, see @ref{Contextual Precedence, , , bison}, in the
258 Is the list of nonterminal definitions. Each definition has the form:
260 @code{(@var{nonterm} . @var{rules})}
262 Where @var{nonterm} is the nonterminal symbol defined and
263 @var{rules} the list of rules that describe this nonterminal. Each
266 @code{(@var{components} [@var{precedence}] [@var{action}])}
272 Is a list of various terminals and nonterminals that are put together
279 (exp ((exp ?+ exp)) ;; exp: exp '+' exp
284 Says that two groupings of type @samp{exp}, with a @samp{+} token in
285 between, can be combined into a larger grouping of type @samp{exp}.
287 @cindex grammar coding conventions
288 By convention, a nonterminal symbol should be in lower case, such as
289 @samp{exp}, @samp{stmt} or @samp{declaration}. Terminal symbols
290 should be upper case to distinguish them from nonterminals: for
291 example, @samp{INTEGER}, @samp{IDENTIFIER}, @samp{IF} or
292 @samp{RETURN}. A terminal symbol that represents a particular keyword
293 in the language is conventionally the same as that keyword converted
294 to upper case. The terminal symbol @code{error} is reserved for error
297 @cindex middle-rule actions
298 Scattered among the components can be @dfn{middle-rule} actions.
299 Usually only @var{action} is provided (@pxref{action}).
301 If @var{components} in a rule is @code{nil}, it means that the rule
302 can match the empty string. For example, here is how to define a
303 comma-separated sequence of zero or more @samp{exp} groupings:
307 (expseq (nil) ;; expseq: ;; empty
308 ((expseq1)) ;; | expseq1
311 (expseq1 ((exp)) ;; expseq1: exp
312 ((expseq1 ?, exp)) ;; | expseq1 ',' exp
317 @cindex precedence level
319 Assign the rule the precedence of the given terminal item, overriding
320 the precedence that would be deduced for it, that is the one of the
321 last terminal in it. Notice that only terminals declared in
322 @var{assocs} have a precedence level. The altered rule precedence
323 then affects how conflicts involving that rule are resolved.
325 @var{precedence} is an optional vector of one terminal item.
327 Here is how @var{precedence} solves the problem of unary minus.
328 First, declare a precedence for a fictitious terminal symbol named
329 @code{UMINUS}. There are no tokens of this type, but the symbol
330 serves to stand for its precedence:
334 ((default-prec t) ;; This is the default
340 Now the precedence of @code{UMINUS} can be used in specific rules:
344 (exp @dots{} ;; exp: @dots{}
345 ((exp ?- exp)) ;; | exp '-' exp
347 ((?- exp) [UMINUS]) ;; | '-' exp %prec UMINUS
353 If you forget to append @code{[UMINUS]} to the rule for unary minus,
354 Wisent silently assumes that minus has its usual precedence. This
355 kind of problem can be tricky to debug, since one typically discovers
356 the mistake only by testing the code.
358 Using @code{(default-prec nil)} declaration makes it easier to
359 discover this kind of problem systematically. It causes rules that
360 lack a @var{precedence} modifier to have no precedence, even if the
361 last terminal symbol mentioned in their components has a declared
364 If @code{(default-prec nil)} is in effect, you must specify
365 @var{precedence} for all rules that participate in precedence conflict
366 resolution. Then you will see any shift/reduce conflict until you
367 tell Wisent how to resolve it, either by changing your grammar or by
368 adding an explicit precedence. This will probably add declarations to
369 the grammar, but it helps to protect against incorrect rule
372 The effect of @code{(default-prec nil)} can be reversed by giving
373 @code{(default-prec t)}, which is the default.
375 For more details, see @ref{Contextual Precedence, , , bison}, in the
378 It is important to understand that @var{assocs} declarations defines
379 associativity but also assign a precedence level to terminals. All
380 terminals declared in the same @code{left}, @code{right} or
381 @code{nonassoc} association get the same precedence level. The
382 precedence level is increased at each new association.
384 On the other hand, @var{precedence} explicitly assign the precedence
385 level of the given terminal to a rule.
387 @cindex semantic actions
388 @item @anchor{action}action
389 An action is an optional Emacs Lisp function call, like this:
393 The result of an action determines the semantic value of a rule.
395 From an implementation standpoint, the function call will be embedded
396 in a lambda expression, and several useful local variables will be
402 Where @var{n} is a positive integer. Like in Bison, the value of
403 @code{$@var{n}} is the semantic value of the @var{n}th element of
404 @var{components}, starting from 1. It can be of any Lisp data
407 @vindex $region@var{n}
409 Where @var{n} is a positive integer. For each @code{$@var{n}}
410 variable defined there is a corresponding @code{$region@var{n}}
411 variable. Its value is a pair @code{(@var{start-pos} .
412 @var{end-pos})} that represent the start and end positions (in the
413 lexical input stream) of the @code{$@var{n}} value. It can be
414 @code{nil} when the component positions are not available, like for an
415 empty string component for example.
419 Its value is the leftmost and rightmost positions of input data
420 matched by all @var{components} in the rule. This is a pair
421 @code{(@var{leftmost-pos} . @var{rightmost-pos})}. It can be
422 @code{nil} when components positions are not available.
426 This variable is initialized with the nonterminal symbol
427 (@var{nonterm}) the rule belongs to. It could be useful to improve
428 error reporting or debugging. It is also used to automatically
429 provide incremental re-parse entry points for @semantic{} tags
430 (@pxref{Wisent Semantic}).
434 The value of @code{$action} is the symbolic name of the current
435 semantic action (@pxref{Debugging actions}).
438 When an action is not specified a default value is supplied, it is
439 @code{(identity $1)}. This means that the default semantic value of a
440 rule is the value of its first component. Excepted for a rule
441 matching the empty string, for which the default action is to return
446 @node Example, Compiling a grammar, Grammar format, Wisent Grammar
447 @comment node-name, next, previous, up
450 @cindex grammar example
451 Here is an example to parse simple infix arithmetic expressions. See
452 @ref{Infix Calc, , , bison}, in the Bison manual for details.
460 ;; Terminal associativity & precedence
471 (format "%s %s" $1 $2))
485 (string-to-number $1))
507 In the bison-like @dfn{WY} format (@pxref{Wisent Semantic}) the
508 grammar looks like this:
514 %nonassoc '=' ;; comparison
517 %left NEG ;; negation--unary minus
518 %right '^' ;; exponentiation
525 (format "%s %s" $1 $2)
539 (string-to-number $1)
562 @node Compiling a grammar, Conflicts, Example, Wisent Grammar
563 @comment node-name, next, previous, up
564 @section Compiling a grammar
567 After providing a context-free grammar in a suitable format, it must
568 be translated into a set of tables (an @dfn{automaton}) that will be
569 used to derive the parser. Like Bison, Wisent translates grammars that
570 must be @dfn{LALR(1)}.
572 @cindex LALR(1) grammar
573 @cindex look-ahead token
574 A grammar is @acronym{LALR(1)} if it is possible to tell how to parse
575 any portion of an input string with just a single token of look-ahead:
576 the @dfn{look-ahead token}. See @ref{Language and Grammar, , ,
577 bison}, in the Bison manual for more information.
579 @cindex grammar compilation
580 Grammar translation (compilation) is achieved by the function:
582 @cindex compiling a grammar
583 @vindex wisent-single-start-flag
584 @findex wisent-compile-grammar
585 @defun wisent-compile-grammar grammar &optional start-list
586 Compile @var{grammar} and return an @acronym{LALR(1)} automaton.
588 Optional argument @var{start-list} is a list of start symbols
589 (nonterminals). If @code{nil} the first nonterminal defined in the
590 grammar is the default start symbol. If @var{start-list} contains
591 only one element, it defines the start symbol. If @var{start-list}
592 contains more than one element, all are defined as potential start
593 symbols, unless @code{wisent-single-start-flag} is non-@code{nil}. In
594 that case the first element of @var{start-list} defines the start
595 symbol and others are ignored.
597 The @acronym{LALR(1)} automaton is a vector of the form:
599 @code{[@var{actions gotos starts functions}]}
603 A state/token matrix telling the parser what to do at every state
604 based on the current look-ahead token. That is shift, reduce, accept
605 or error. See also @ref{Wisent Parsing}.
608 A state/nonterminal matrix telling the parser the next state to go to
609 after reducing with each rule.
612 An alist which maps the allowed start symbols (nonterminals) to
613 lexical tokens that will be first shifted into the parser stack.
616 An obarray of semantic action symbols. A semantic action is actually
617 an Emacs Lisp function (lambda expression).
621 @node Conflicts, , Compiling a grammar, Wisent Grammar
622 @comment node-name, next, previous, up
625 Normally, a grammar should produce an automaton where at each state
626 the parser has only one action to do (@pxref{Wisent Parsing}).
628 @cindex ambiguous grammar
629 In certain cases, a grammar can produce an automaton where, at some
630 states, there are more than one action possible. Such a grammar is
631 @dfn{ambiguous}, and generates @dfn{conflicts}.
633 @cindex deterministic automaton
634 The parser can't be driven by an automaton which isn't completely
635 @dfn{deterministic}, that is which contains conflicts. It is
636 necessary to resolve the conflicts to eliminate them. Wisent resolves
637 conflicts like Bison does.
639 @cindex grammar conflicts
640 @cindex conflicts resolution
641 There are two sorts of conflicts:
644 @cindex shift/reduce conflicts
645 @item shift/reduce conflicts
646 When either a shift or a reduction would be valid at the same state.
648 Such conflicts are resolved by choosing to shift, unless otherwise
649 directed by operator precedence declarations.
650 See @ref{Shift/Reduce , , , bison}, in the Bison manual for more
653 @cindex reduce/reduce conflicts
654 @item reduce/reduce conflicts
655 That occurs if there are two or more rules that apply to the same
656 sequence of input. This usually indicates a serious error in the
659 Such conflicts are resolved by choosing to use the rule that appears
660 first in the grammar, but it is very risky to rely on this. Every
661 reduce/reduce conflict must be studied and usually eliminated. See
662 @ref{Reduce/Reduce , , , bison}, in the Bison manual for more
667 * Grammar Debugging::
668 * Understanding the automaton::
671 @node Grammar Debugging
672 @subsection Grammar debugging
674 @cindex grammar debugging
675 @cindex grammar verbose description
676 To help writing a new grammar, @code{wisent-compile-grammar} can
677 produce a verbose report containing a detailed description of the
678 grammar and parser (equivalent to what Bison reports with the
679 @option{--verbose} option).
681 To enable the verbose report you can set to non-@code{nil} the
684 @vindex wisent-verbose-flag
685 @deffn Option wisent-verbose-flag
686 non-@code{nil} means to report verbose information on generated parser.
689 Or interactively use the command:
691 @findex wisent-toggle-verbose-flag
692 @deffn Command wisent-toggle-verbose-flag
693 Toggle whether to report verbose information on generated parser.
696 The verbose report is printed in the temporary buffer
697 @code{*wisent-log*} when running interactively, or in file
698 @file{wisent.output} when running in batch mode. Different
699 reports are separated from each other by a line like this:
703 *** Wisent @var{source-file} - 2002-06-27 17:33
707 where @var{source-file} is the name of the Emacs Lisp file from which
708 the grammar was read. See @ref{Understanding the automaton}, for
709 details on the verbose report.
713 To help debugging the grammar compiler itself, you can set this
714 variable to print the content of some internal data structures:
716 @vindex wisent-debug-flag
717 @defvar wisent-debug-flag
718 non-@code{nil} means enable some debug stuff.
722 @node Understanding the automaton
723 @subsection Understanding the automaton
725 @cindex understanding the automaton
726 This section (took from the manual of Bison 1.49) describes how to use
727 the verbose report printed by @code{wisent-compile-grammar} to
728 understand the generated automaton, to tune or fix a grammar.
730 We will use the following example:
734 (let ((wisent-verbose-flag t)) ;; Print a verbose report!
735 (wisent-compile-grammar
736 '((NUM STR) ; %token NUM STR
738 ((left ?+ ?-) ; %left '+' '-';
739 (left ?*)) ; %left '*'
742 ((exp ?+ exp)) ; exp '+' exp
743 ((exp ?- exp)) ; | exp '-' exp
744 ((exp ?* exp)) ; | exp '*' exp
745 ((exp ?/ exp)) ; | exp '/' exp
753 'nil) ; no %start declarations
758 When evaluating the above expression, grammar compilation first issues
759 the following two clear messages:
763 Grammar contains 1 useless nonterminals and 1 useless rules
764 Grammar contains 7 shift/reduce conflicts
768 The @samp{*wisent-log*} buffer details things!
770 The first section reports conflicts that were solved using precedence
771 and/or associativity:
775 Conflict in state 7 between rule 1 and token '+' resolved as reduce.
776 Conflict in state 7 between rule 1 and token '-' resolved as reduce.
777 Conflict in state 7 between rule 1 and token '*' resolved as shift.
778 Conflict in state 8 between rule 2 and token '+' resolved as reduce.
779 Conflict in state 8 between rule 2 and token '-' resolved as reduce.
780 Conflict in state 8 between rule 2 and token '*' resolved as shift.
781 Conflict in state 9 between rule 3 and token '+' resolved as reduce.
782 Conflict in state 9 between rule 3 and token '-' resolved as reduce.
783 Conflict in state 9 between rule 3 and token '*' resolved as reduce.
787 The next section reports useless tokens, nonterminal and rules (note
788 that useless tokens might be used by the scanner):
792 Useless nonterminals:
797 Terminals which are not used:
808 The next section lists states that still have conflicts:
812 State 7 contains 1 shift/reduce conflict.
813 State 8 contains 1 shift/reduce conflict.
814 State 9 contains 1 shift/reduce conflict.
815 State 10 contains 4 shift/reduce conflicts.
819 The next section reproduces the grammar used:
834 And reports the uses of the symbols:
838 Terminals, with rules where they appear
850 Nonterminals, with rules where they appear
853 on left: 1 2 3 4 5, on right: 1 2 3 4
857 The report then details the automaton itself, describing each state
858 with it set of @dfn{items}, also known as @dfn{pointed rules}. Each
859 item is a production rule together with a point (marked by @samp{.})
860 that the input cursor.
866 NUM shift, and go to state 1
872 State 0 corresponds to being at the very beginning of the parsing, in
873 the initial rule, right before the start symbol (@samp{exp}). When
874 the parser returns to this state right after having reduced a rule
875 that produced an @samp{exp}, it jumps to state 2. If there is no such
876 transition on a nonterminal symbol, and the lookahead is a @samp{NUM},
877 then this token is shifted on the parse stack, and the control flow
878 jumps to state 1. Any other lookahead triggers a parse error.
886 exp -> NUM . (rule 5)
888 $default reduce using rule 5 (exp)
892 the rule 5, @samp{exp: NUM;}, is completed. Whatever the lookahead
893 (@samp{$default}), the parser will reduce it. If it was coming from
894 state 0, then, after this reduction it will return to state 0, and
895 will jump to state 2 (@samp{exp: go to state 2}).
901 exp -> exp . '+' exp (rule 1)
902 exp -> exp . '-' exp (rule 2)
903 exp -> exp . '*' exp (rule 3)
904 exp -> exp . '/' exp (rule 4)
906 $EOI shift, and go to state 11
907 '+' shift, and go to state 3
908 '-' shift, and go to state 4
909 '*' shift, and go to state 5
910 '/' shift, and go to state 6
914 In state 2, the automaton can only shift a symbol. For instance,
915 because of the item @samp{exp -> exp . '+' exp}, if the lookahead if
916 @samp{+}, it will be shifted on the parse stack, and the automaton
917 control will jump to state 3, corresponding to the item
918 @samp{exp -> exp . '+' exp}:
924 exp -> exp '+' . exp (rule 1)
926 NUM shift, and go to state 1
932 Since there is no default action, any other token than those listed
933 above will trigger a parse error.
935 The interpretation of states 4 to 6 is straightforward:
941 exp -> exp '-' . exp (rule 2)
943 NUM shift, and go to state 1
951 exp -> exp '*' . exp (rule 3)
953 NUM shift, and go to state 1
961 exp -> exp '/' . exp (rule 4)
963 NUM shift, and go to state 1
969 As was announced in beginning of the report, @samp{State 7 contains 1
970 shift/reduce conflict.}:
976 exp -> exp . '+' exp (rule 1)
977 exp -> exp '+' exp . (rule 1)
978 exp -> exp . '-' exp (rule 2)
979 exp -> exp . '*' exp (rule 3)
980 exp -> exp . '/' exp (rule 4)
982 '*' shift, and go to state 5
983 '/' shift, and go to state 6
985 '/' [reduce using rule 1 (exp)]
986 $default reduce using rule 1 (exp)
990 Indeed, there are two actions associated to the lookahead @samp{/}:
991 either shifting (and going to state 6), or reducing rule 1. The
992 conflict means that either the grammar is ambiguous, or the parser
993 lacks information to make the right decision. Indeed the grammar is
994 ambiguous, as, since we did not specify the precedence of @samp{/},
995 the sentence @samp{NUM + NUM / NUM} can be parsed as @samp{NUM + (NUM
996 / NUM)}, which corresponds to shifting @samp{/}, or as @samp{(NUM +
997 NUM) / NUM}, which corresponds to reducing rule 1.
999 Because in @acronym{LALR(1)} parsing a single decision can be made,
1000 Wisent arbitrarily chose to disable the reduction, see
1001 @ref{Conflicts}. Discarded actions are reported in between square
1004 Note that all the previous states had a single possible action: either
1005 shifting the next token and going to the corresponding state, or
1006 reducing a single rule. In the other cases, i.e., when shifting
1007 @emph{and} reducing is possible or when @emph{several} reductions are
1008 possible, the lookahead is required to select the action. State 7 is
1009 one such state: if the lookahead is @samp{*} or @samp{/} then the
1010 action is shifting, otherwise the action is reducing rule 1. In other
1011 words, the first two items, corresponding to rule 1, are not eligible
1012 when the lookahead is @samp{*}, since we specified that @samp{*} has
1013 higher precedence that @samp{+}. More generally, some items are
1014 eligible only with some set of possible lookaheads.
1016 States 8 to 10 are similar:
1022 exp -> exp . '+' exp (rule 1)
1023 exp -> exp . '-' exp (rule 2)
1024 exp -> exp '-' exp . (rule 2)
1025 exp -> exp . '*' exp (rule 3)
1026 exp -> exp . '/' exp (rule 4)
1028 '*' shift, and go to state 5
1029 '/' shift, and go to state 6
1031 '/' [reduce using rule 2 (exp)]
1032 $default reduce using rule 2 (exp)
1037 exp -> exp . '+' exp (rule 1)
1038 exp -> exp . '-' exp (rule 2)
1039 exp -> exp . '*' exp (rule 3)
1040 exp -> exp '*' exp . (rule 3)
1041 exp -> exp . '/' exp (rule 4)
1043 '/' shift, and go to state 6
1045 '/' [reduce using rule 3 (exp)]
1046 $default reduce using rule 3 (exp)
1051 exp -> exp . '+' exp (rule 1)
1052 exp -> exp . '-' exp (rule 2)
1053 exp -> exp . '*' exp (rule 3)
1054 exp -> exp . '/' exp (rule 4)
1055 exp -> exp '/' exp . (rule 4)
1057 '+' shift, and go to state 3
1058 '-' shift, and go to state 4
1059 '*' shift, and go to state 5
1060 '/' shift, and go to state 6
1062 '+' [reduce using rule 4 (exp)]
1063 '-' [reduce using rule 4 (exp)]
1064 '*' [reduce using rule 4 (exp)]
1065 '/' [reduce using rule 4 (exp)]
1066 $default reduce using rule 4 (exp)
1070 Observe that state 10 contains conflicts due to the lack of precedence
1071 of @samp{/} wrt @samp{+}, @samp{-}, and @samp{*}, but also because the
1072 associativity of @samp{/} is not specified.
1074 Finally, the state 11 (plus 12) is named the @dfn{final state}, or the
1075 @dfn{accepting state}:
1081 $EOI shift, and go to state 12
1091 The end of input is shifted @samp{$EOI shift,} and the parser exits
1092 successfully (@samp{go to state 12}, that terminates).
1094 @node Wisent Parsing
1095 @chapter Wisent Parsing
1097 @cindex bottom-up parser
1098 @cindex shift-reduce parser
1099 The Wisent's parser is what is called a @dfn{bottom-up} or
1100 @dfn{shift-reduce} parser which repeatedly:
1105 That is pushes the value of the last lexical token read (the
1106 look-ahead token) into a value stack, and reads a new one.
1110 That is replaces a nonterminal by its semantic value. The values of
1111 the components which form the right hand side of a rule are popped
1112 from the value stack and reduced by the semantic action of this rule.
1113 The result is pushed back on top of value stack.
1116 The parser will stop on:
1121 When all input has been successfully parsed. The semantic value of
1122 the start nonterminal is on top of the value stack.
1124 @cindex syntax error
1126 When a syntax error (an unexpected token in input) has been detected.
1127 At this point the parser issues an error message and either stops or
1128 calls a recovery routine to try to resume parsing.
1131 @cindex table-driven parser
1132 The above elementary actions are driven by the @acronym{LALR(1)}
1133 automaton built by @code{wisent-compile-grammar} from a context-free
1136 The Wisent's parser is entered by calling the function:
1138 @findex wisent-parse
1139 @defun wisent-parse automaton lexer &optional error start
1140 Parse input using the automaton specified in @var{automaton}.
1144 Is an @acronym{LALR(1)} automaton generated by
1145 @code{wisent-compile-grammar} (@pxref{Wisent Grammar}).
1148 Is a function with no argument called by the parser to obtain the next
1149 terminal (token) in input (@pxref{Writing a lexer}).
1152 Is an optional reporting function called when a parse error occurs.
1153 It receives a message string to report. It defaults to the function
1154 @code{wisent-message} (@pxref{Report errors}).
1157 Specify the start symbol (nonterminal) used by the parser as its goal.
1158 It defaults to the start symbol defined in the grammar
1159 (@pxref{Wisent Grammar}).
1163 The following two normal hooks permit to do some useful processing
1164 respectively before to start parsing, and after the parser terminated.
1166 @vindex wisent-pre-parse-hook
1167 @defvar wisent-pre-parse-hook
1168 Normal hook run just before entering the @var{LR} parser engine.
1171 @vindex wisent-post-parse-hook
1172 @defvar wisent-post-parse-hook
1173 Normal hook run just after the @var{LR} parser engine terminated.
1181 * Debugging actions::
1184 @node Writing a lexer
1185 @section What the parser must receive
1187 It is important to understand that the parser does not parse
1188 characters, but lexical tokens, and does not know anything about
1189 characters in text streams!
1191 @cindex lexical analysis
1194 Reading input data to produce lexical tokens is performed by a lexer
1195 (also called a scanner) in a lexical analysis step, before the syntax
1196 analysis step performed by the parser. The parser automatically calls
1197 the lexer when it needs the next token to parse.
1199 @cindex lexical tokens
1200 A Wisent's lexer is an Emacs Lisp function with no argument. It must
1201 return a valid lexical token of the form:
1203 @code{(@var{token-class value} [@var{start} . @var{end}])}
1207 Is a category of lexical token identifying a terminal as specified in
1208 the grammar (@pxref{Wisent Grammar}). It can be a symbol or a character
1212 Is the value of the lexical token. It can be of any valid Emacs Lisp
1217 Are the optionals beginning and end positions of @var{value} in the
1221 When there are no more tokens to read the lexer must return the token
1222 @code{(list wisent-eoi-term)} to each request.
1224 @vindex wisent-eoi-term
1225 @defvar wisent-eoi-term
1226 Predefined constant, End-Of-Input terminal symbol.
1229 @code{wisent-lex} is an example of a lexer that reads lexical tokens
1230 produced by a @semantic{} lexer, and translates them into lexical tokens
1231 suitable to the Wisent parser. See also @ref{Wisent Lex}.
1233 To call the lexer in a semantic action use the function
1234 @code{wisent-lexer}. See also @ref{Actions goodies}.
1236 @node Actions goodies
1237 @section Variables and macros useful in grammar actions.
1239 @vindex wisent-input
1240 @defvar wisent-input
1241 The last token read.
1242 This variable only has meaning in the scope of @code{wisent-parse}.
1245 @findex wisent-lexer
1247 Obtain the next terminal in input.
1250 @findex wisent-region
1251 @defun wisent-region &rest positions
1252 Return the start/end positions of the region including
1253 @var{positions}. Each element of @var{positions} is a pair
1254 @w{@code{(@var{start-pos} . @var{end-pos})}} or @code{nil}. The
1255 returned value is the pair @w{@code{(@var{min-start-pos} .
1256 @var{max-end-pos})}} or @code{nil} if no @var{positions} are
1261 @section The error reporting function
1263 @cindex error reporting
1264 When the parser encounters a syntax error it calls a user-defined
1265 function. It must be an Emacs Lisp function with one argument: a
1266 string containing the message to report.
1268 By default the parser uses this function to report error messages:
1270 @findex wisent-message
1271 @defun wisent-message string &rest args
1272 Print a one-line message if @code{wisent-parse-verbose-flag} is set.
1273 Pass @var{string} and @var{args} arguments to @dfn{message}.
1278 @code{wisent-message} uses the following function to print lexical
1281 @defun wisent-token-to-string token
1282 Return a printed representation of lexical token @var{token}.
1285 The general printed form of a lexical token is:
1287 @w{@code{@var{token}(@var{value})@@@var{location}}}
1290 To control the verbosity of the parser you can set to non-@code{nil}
1293 @vindex wisent-parse-verbose-flag
1294 @deffn Option wisent-parse-verbose-flag
1295 non-@code{nil} means to issue more messages while parsing.
1298 Or interactively use the command:
1300 @findex wisent-parse-toggle-verbose-flag
1301 @deffn Command wisent-parse-toggle-verbose-flag
1302 Toggle whether to issue more messages while parsing.
1305 When the error reporting function is entered the variable
1306 @code{wisent-input} contains the unexpected token as returned by the
1309 The error reporting function can be called from a semantic action too
1310 using the special macro @code{wisent-error}. When called from a
1311 semantic action entered by error recovery (@pxref{Error recovery}) the
1312 value of the variable @code{wisent-recovering} is non-@code{nil}.
1314 @node Error recovery
1315 @section Error recovery
1317 @cindex error recovery
1318 The error recovery mechanism of the Wisent's parser conforms to the
1319 one Bison uses. See @ref{Error Recovery, , , bison}, in the Bison
1323 To recover from a syntax error you must write rules to recognize the
1324 special token @code{error}. This is a terminal symbol that is
1325 automatically defined and reserved for error handling.
1327 When the parser encounters a syntax error, it pops the state stack
1328 until it finds a state that allows shifting the @code{error} token.
1329 After it has been shifted, if the old look-ahead token is not
1330 acceptable to be shifted next, the parser reads tokens and discards
1331 them until it finds a token which is acceptable.
1333 @cindex error recovery strategy
1334 Strategies for error recovery depend on the choice of error rules in
1335 the grammar. A simple and useful strategy is simply to skip the rest
1336 of the current statement if an error is detected:
1340 (stmnt (( error ?; )) ;; on error, skip until ';' is read
1345 It is also useful to recover to the matching close-delimiter of an
1346 opening-delimiter that has already been parsed:
1350 (primary (( ?@{ expr ?@} ))
1357 @cindex error recovery actions
1358 Note that error recovery rules may have actions, just as any other
1359 rules can. Here are some predefined hooks, variables, functions or
1360 macros, useful in such actions:
1362 @vindex wisent-nerrs
1363 @defvar wisent-nerrs
1364 The number of parse errors encountered so far.
1367 @vindex wisent-recovering
1368 @defvar wisent-recovering
1369 non-@code{nil} means that the parser is recovering.
1370 This variable only has meaning in the scope of @code{wisent-parse}.
1373 @findex wisent-error
1374 @defun wisent-error msg
1375 Call the user supplied error reporting function with message
1376 @var{msg} (@pxref{Report errors}).
1378 For an example of use, @xref{wisent-skip-token}.
1381 @findex wisent-errok
1383 Resume generating error messages immediately for subsequent syntax
1386 The parser suppress error message for syntax errors that happens
1387 shortly after the first, until three consecutive input tokens have
1388 been successfully shifted.
1390 Calling @code{wisent-errok} in an action, make error messages resume
1391 immediately. No error messages will be suppressed if you call it in
1392 an error rule's action.
1394 For an example of use, @xref{wisent-skip-token}.
1397 @findex wisent-clearin
1398 @defun wisent-clearin
1399 Discard the current lookahead token.
1400 This will cause a new lexical token to be read.
1402 In an error rule's action the previous lookahead token is reanalyzed
1403 immediately. @code{wisent-clearin} may be called to clear this token.
1405 For example, suppose that on a parse error, an error handling routine
1406 is called that advances the input stream to some point where parsing
1407 should once again commence. The next symbol returned by the lexical
1408 scanner is probably correct. The previous lookahead token ought to
1409 be discarded with @code{wisent-clearin}.
1411 For an example of use, @xref{wisent-skip-token}.
1414 @findex wisent-abort
1416 Abort parsing and save the lookahead token.
1419 @findex wisent-set-region
1420 @defun wisent-set-region start end
1421 Change the region of text matched by the current nonterminal.
1422 @var{start} and @var{end} are respectively the beginning and end
1423 positions of the region occupied by the group of components associated
1424 to this nonterminal. If @var{start} or @var{end} values are not a
1425 valid positions the region is set to @code{nil}.
1427 For an example of use, @xref{wisent-skip-token}.
1430 @vindex wisent-discarding-token-functions
1431 @defvar wisent-discarding-token-functions
1432 List of functions to be called when discarding a lexical token.
1433 These functions receive the lexical token discarded.
1434 When the parser encounters unexpected tokens, it can discards them,
1435 based on what directed by error recovery rules. Either when the
1436 parser reads tokens until one is found that can be shifted, or when an
1437 semantic action calls the function @code{wisent-skip-token} or
1438 @code{wisent-skip-block}.
1439 For language specific hooks, make sure you define this as a local
1442 For example, in @semantic{}, this hook is set to the function
1443 @code{wisent-collect-unmatched-syntax} to collect unmatched lexical
1444 tokens (@pxref{Useful functions}).
1447 @findex wisent-skip-token
1448 @defun wisent-skip-token
1449 @anchor{wisent-skip-token}
1450 Skip the lookahead token in order to resume parsing.
1452 Must be used in error recovery semantic actions.
1454 It typically looks like this:
1458 (wisent-message "%s: skip %s" $action
1459 (wisent-token-to-string wisent-input))
1461 'wisent-discarding-token-functions wisent-input)
1468 @findex wisent-skip-block
1469 @defun wisent-skip-block
1470 Safely skip a block in order to resume parsing.
1472 Must be used in error recovery semantic actions.
1474 A block is data between an open-delimiter (syntax class @code{(}) and
1475 a matching close-delimiter (syntax class @code{)}):
1479 (a parenthesized block)
1480 [a block between brackets]
1481 @{a block between braces@}
1485 The following example uses @code{wisent-skip-block} to safely skip a
1486 block delimited by @samp{LBRACE} (@code{@{}) and @samp{RBRACE}
1487 (@code{@}}) tokens, when a syntax error occurs in
1488 @samp{other-components}:
1492 (block ((LBRACE other-components RBRACE))
1495 (wisent-skip-block))
1501 @node Debugging actions
1502 @section Debugging semantic actions
1504 @cindex semantic action symbols
1505 Each semantic action is represented by a symbol interned in an
1506 @dfn{obarray} that is part of the @acronym{LALR(1)} automaton
1507 (@pxref{Compiling a grammar}). @code{symbol-function} on a semantic
1508 action symbol return the semantic action lambda expression.
1510 A semantic action symbol name has the form
1511 @code{@var{nonterminal}:@var{index}}, where @var{nonterminal} is the
1512 name of the nonterminal symbol the action belongs to, and @var{index}
1513 is an action sequence number within the scope of @var{nonterminal}.
1514 For example, this nonterminal definition:
1519 line [@code{input:0}]
1521 (format "%s %s" $1 $2) [@code{input:1}]
1526 Will produce two semantic actions, and associated symbols:
1530 A default action that returns @code{$1}.
1533 That returns @code{(format "%s %s" $1 $2)}.
1536 @cindex debugging semantic actions
1537 Debugging uses the Lisp debugger to investigate what is happening
1538 during execution of semantic actions.
1539 Three commands are available to debug semantic actions. They receive
1543 @item The automaton that contains the semantic action.
1545 @item The semantic action symbol.
1548 @findex wisent-debug-on-entry
1549 @deffn Command wisent-debug-on-entry automaton function
1550 Request @var{automaton}'s @var{function} to invoke debugger each time it is called.
1551 @var{function} must be a semantic action symbol that exists in @var{automaton}.
1554 @findex wisent-cancel-debug-on-entry
1555 @deffn Command wisent-cancel-debug-on-entry automaton function
1556 Undo effect of @code{wisent-debug-on-entry} on @var{automaton}'s @var{function}.
1557 @var{function} must be a semantic action symbol that exists in @var{automaton}.
1560 @findex wisent-debug-show-entry
1561 @deffn Command wisent-debug-show-entry automaton function
1562 Show the source of @var{automaton}'s semantic action @var{function}.
1563 @var{function} must be a semantic action symbol that exists in @var{automaton}.
1566 @node Wisent Semantic
1567 @chapter How to use Wisent with Semantic
1570 This section presents how the Wisent's parser can be used to produce
1571 @dfn{tags} for the @semantic{} tool set.
1573 @semantic{} tags form a hierarchy of Emacs Lisp data structures that
1574 describes a program in a way independent of programming languages.
1575 Tags map program declarations, like functions, methods, variables,
1576 data types, classes, includes, grammar rules, etc..
1578 @cindex WY grammar format
1579 To use the Wisent parser with @semantic{} you have to define
1580 your grammar in @dfn{WY} form, a grammar format very close
1581 to the one used by Bison.
1583 Please @inforef{top, Semantic Grammar Framework Manual, grammar-fw}
1584 for more information on @semantic{} grammars.
1591 @node Grammar styles
1592 @section Grammar styles
1594 @cindex grammar styles
1595 @semantic{} parsing heavily depends on how you wrote the grammar.
1596 There are mainly two styles to write a Wisent's grammar intended to be
1597 used with the @semantic{} tool set: the @dfn{Iterative style} and the
1598 @dfn{Bison style}. Each one has pros and cons, and in certain cases
1599 it can be worth a mix of the two styles!
1605 * Start nonterminals::
1606 * Useful functions::
1609 @node Iterative style, Bison style, Grammar styles, Grammar styles
1610 @subsection Iterative style
1612 @cindex grammar iterative style
1613 The @dfn{iterative style} is the preferred style to use with @semantic{}.
1614 It relies on an iterative parser back-end mechanism which parses start
1615 nonterminals one at a time and automagically skips unexpected lexical
1618 Compared to rule-based iterative functions (@pxref{Bison style}),
1619 iterative parsers are better in that they can handle obscure errors
1623 Each start nonterminal must produces a @dfn{raw tag} by calling a
1624 @code{TAG}-like grammar macro with appropriate parameters. See also
1625 @ref{Start nonterminals}.
1627 @cindex expanded tag
1628 Then, each parsing iteration automatically translates a raw tag into
1629 @dfn{expanded tags}, updating the raw tag structure with internal
1630 properties and buffer related data.
1632 After parsing completes, it results in a tree of expanded tags.
1634 The following example is a snippet of the iterative style Java grammar
1635 provided in the @semantic{} distribution in the file
1636 @file{semantic/wisent/java-tags.wy}.
1641 ;; Alternate entry points
1642 ;; - Needed by partial re-parse
1643 %start formal_parameter
1645 ;; - Needed by EXPANDFULL clauses
1646 %start formal_parameters
1649 formal_parameter_list
1651 (EXPANDFULL $1 formal_parameters)
1659 | formal_parameter COMMA
1660 | formal_parameter RPAREN
1664 : formal_parameter_modifier_opt type variable_declarator_id
1665 (VARIABLE-TAG $3 $2 nil :typemodifiers $1)
1671 It shows the use of the @code{EXPANDFULL} grammar macro to parse a
1672 @samp{PAREN_BLOCK} which contains a @samp{formal_parameter_list}.
1673 @code{EXPANDFULL} tells to recursively parse @samp{formal_parameters}
1674 inside @samp{PAREN_BLOCK}. The parser iterates until it digested all
1675 available input data inside the @samp{PAREN_BLOCK}, trying to match
1676 any of the @samp{formal_parameters} rules:
1683 @item @samp{formal_parameter COMMA}
1685 @item @samp{formal_parameter RPAREN}
1688 At each iteration it will return a @samp{formal_parameter} raw tag,
1689 or @code{nil} to skip unwanted (single @samp{LPAREN} or @samp{RPAREN}
1690 for example) or unexpected input data. Those raw tags will be
1691 automatically expanded by the iterative back-end parser.
1694 @subsection Bison style
1696 @cindex grammar bison style
1697 What we call the @dfn{Bison style} is the traditional style of Bison's
1698 grammars. Compared to iterative style, it is not straightforward to
1699 use grammars written in Bison style in @semantic{}. Mainly because such
1700 grammars are designed to parse the whole input data in one pass, and
1701 don't use the iterative parser back-end mechanism (@pxref{Iterative
1702 style}). With Bison style the parser is called once to parse the
1703 grammar start nonterminal.
1705 The following example is a snippet of the Bison style Java grammar
1706 provided in the @semantic{} distribution in the file
1707 @file{semantic/wisent/java.wy}.
1711 %start formal_parameter
1714 formal_parameter_list
1715 : formal_parameter_list COMMA formal_parameter
1722 : formal_parameter_modifier_opt type variable_declarator_id
1724 (VARIABLE-TAG $3 $2 :typemodifiers $1)
1730 The first consequence is that syntax errors are not automatically
1731 handled by @semantic{}. Thus, it is necessary to explicitly handle
1732 them at the grammar level, providing error recovery rules to skip
1733 unexpected input data.
1735 The second consequence is that the iterative parser can't do automatic
1736 tag expansion, except for the start nonterminal value. It is
1737 necessary to explicitly expand tags from concerned semantic actions by
1738 calling the grammar macro @code{EXPANDTAG} with a raw tag as
1739 parameter. See also @ref{Start nonterminals}, for incremental
1740 re-parse considerations.
1743 @subsection Mixed style
1745 @cindex grammar mixed style
1750 %start prologue epilogue declaration nonterminal rule
1765 SYMBOL COLON rules SEMI
1766 (TAG $1 'nonterminal :children $3)
1771 (apply 'nconc (nreverse $1))
1784 name type comps prec action elt)
1787 (TAG name 'rule :type type :value comps :prec prec :expr action)
1793 This example shows how iterative and Bison styles can be combined in
1794 the same grammar to obtain a good compromise between grammar
1795 complexity and an efficient parsing strategy in an interactive
1798 @samp{nonterminal} is parsed using iterative style via the main
1799 @samp{grammar} rule. The semantic action uses the @code{TAG} macro to
1800 produce a raw tag, automagically expanded by @semantic{}.
1802 But @samp{rules} part is parsed in Bison style! Why?
1804 Rule delimiters are the colon (@code{:}), that follows the nonterminal
1805 name, and a final semicolon (@code{;}). Unfortunately these
1806 delimiters are not @code{open-paren}/@code{close-paren} type, and the
1807 Emacs' syntactic analyzer can't easily isolate data between them to
1808 produce a @samp{RULES_PART} parenthesis-block-like lexical token.
1809 Consequently it is not possible to use @code{EXPANDFULL} to iterate in
1810 @samp{RULES_PART}, like this:
1815 SYMBOL COLON rules SEMI
1816 (TAG $1 'nonterminal :children $3)
1820 RULES_PART ;; @strong{Map a parenthesis-block-like lexical token}
1821 (EXPANDFULL $1 'rules)
1834 name type comps prec action elt)
1836 (TAG name 'rule :type type :value comps :prec prec :expr action)
1842 In such cases, when it is difficult for Emacs to obtain
1843 parenthesis-block-like lexical tokens, the best solution is to use the
1844 traditional Bison style with error recovery!
1846 In some extreme cases, it can also be convenient to extend the lexer,
1847 to deliver new lexical tokens, to simplify the grammar.
1849 @node Start nonterminals
1850 @subsection Start nonterminals
1852 @cindex start nonterminals
1853 @cindex @code{reparse-symbol} property
1854 When you write a grammar for @semantic{}, it is important to carefully
1855 indicate the start nonterminals. Each one defines an entry point in
1856 the grammar, and after parsing its semantic value is returned to the
1857 back-end iterative engine. Consequently:
1859 @strong{The semantic value of a start nonterminal must be a produced
1860 by a TAG like grammar macro}.
1862 Start nonterminals are declared by @code{%start} statements. When
1863 nothing is specified the first nonterminal that appears in the grammar
1864 is the start nonterminal.
1866 Generally, the following nonterminals must be declared as start
1870 @item The main grammar entry point
1875 @item nonterminals passed to @code{EXPAND}/@code{EXPANDFULL}
1877 These grammar macros recursively parse a part of input data, based on
1878 rules of the given nonterminal.
1880 For example, the following will parse @samp{PAREN_BLOCK} data using
1881 the @samp{formal_parameters} rules:
1885 formal_parameter_list
1887 (EXPANDFULL $1 formal_parameters)
1892 The semantic value of @samp{formal_parameters} becomes the value of
1893 the @code{EXPANDFULL} expression. It is a list of @semantic{} tags
1894 spliced in the tags tree.
1896 Because the automaton must know that @samp{formal_parameters} is a
1897 start symbol, you must declare it like this:
1901 %start formal_parameters
1907 @cindex incremental re-parse
1908 @cindex reparse-symbol
1909 The @code{EXPANDFULL} macro has a side effect it is important to know,
1910 related to the incremental re-parse mechanism of @semantic{}: the
1911 nonterminal symbol parameter passed to @code{EXPANDFULL} also becomes
1912 the @code{reparse-symbol} property of the tag returned by the
1913 @code{EXPANDFULL} expression.
1915 When buffer's data mapped by a tag is modified, @semantic{}
1916 schedules an incremental re-parse of that data, using the tag's
1917 @code{reparse-symbol} property as start nonterminal.
1919 @strong{The rules associated to such start symbols must be carefully
1920 reviewed to ensure that the incremental parser will work!}
1922 Things are a little bit different when the grammar is written in Bison
1925 @strong{The @code{reparse-symbol} property is set to the nonterminal
1926 symbol the rule that explicitly uses @code{EXPANDTAG} belongs to.}
1935 name type comps prec action elt)
1938 (TAG name 'rule :type type :value comps :prec prec :expr action)
1944 Set the @code{reparse-symbol} property of the expanded tag to
1945 @samp{rule}. A important consequence is that:
1947 @strong{Every nonterminal having any rule that calls @code{EXPANDTAG}
1948 in a semantic action, should be declared as a start symbol!}
1950 @node Useful functions
1951 @subsection Useful functions
1953 Here is a description of some predefined functions it might be useful
1954 to know when writing new code to use Wisent in @semantic{}:
1956 @findex wisent-collect-unmatched-syntax
1957 @defun wisent-collect-unmatched-syntax input
1958 Add @var{input} lexical token to the cache of unmatched tokens, in
1959 variable @code{semantic-unmatched-syntax-cache}.
1961 See implementation of the function @code{wisent-skip-token} in
1962 @ref{Error recovery}, for an example of use.
1966 @section The Wisent Lex lexer
1968 @findex semantic-lex
1969 The lexical analysis step of @semantic{} is performed by the general
1970 function @code{semantic-lex}. For more information, @inforef{Writing
1971 Lexers, ,semantic-langdev}.
1973 @code{semantic-lex} produces lexical tokens of the form:
1977 @code{(@var{token-class start} . @var{end})}
1983 Is a symbol that identifies a lexical token class, like @code{symbol},
1984 @code{string}, @code{number}, or @code{PAREN_BLOCK}.
1988 Are the start and end positions of mapped data in the input buffer.
1991 The Wisent's parser doesn't depend on the nature of analyzed input
1992 stream (buffer, string, etc.), and requires that lexical tokens have a
1993 different form (@pxref{Writing a lexer}):
1997 @code{(@var{token-class value} [@var{start} . @var{end}])}
2001 @cindex lexical token mapping
2002 @code{wisent-lex} is the default Wisent's lexer used in @semantic{}.
2004 @vindex wisent-lex-istream
2007 Return the next available lexical token in Wisent's form.
2009 The variable @code{wisent-lex-istream} contains the list of lexical
2010 tokens produced by @code{semantic-lex}. Pop the next token available
2011 and convert it to a form suitable for the Wisent's parser.
2014 Mapping of lexical tokens as produced by @code{semantic-lex} into
2015 equivalent Wisent lexical tokens is straightforward:
2019 (@var{token-class start} . @var{end})
2020 @result{} (@var{token-class value start} . @var{end})
2024 @var{value} is the input @code{buffer-substring} from @var{start} to
2027 @node GNU Free Documentation License
2028 @appendix GNU Free Documentation License
2030 @include doclicense.texi
2043 @c Following comments are for the benefit of ispell.
2045 @c LocalWords: Wisent automagically wisent Wisent's LALR obarray