FYI: Fix uninitialized location when reporting ambiguity

[Joel E. Denny]

2006-03-06  Joel E. Denny  <address@hidden>

        * tests/glr-regression.at (Uninitialized location when reporting
        ambiguity): New test case.
        * data/glr.c (yyresolveLocations): New function, which uses
        YYLLOC_DEFAULT.
        (yyresolveValue): Invoke yyresolveLocations before reporting an
        ambiguity.
        * doc/bison.texinfo (Default Action for Locations): Note
        YYLLOC_DEFAULT's usage for ambiguity locations.
        (GLR Semantic Actions): Cross-reference those notes.

Index: data/glr.c
===================================================================
RCS file: /sources/bison/bison/data/glr.c,v
retrieving revision 1.167
diff -p -u -r1.167 glr.c
--- data/glr.c  4 Mar 2006 03:29:03 -0000       1.167
+++ data/glr.c  6 Mar 2006 07:35:19 -0000
@@ -1803,6 +1803,57 @@ yyreportAmbiguity (yySemanticOption* yyx
   return yyabort;
 }
 
+/** Starting at and including state S, resolve the location for each of the
+ *  previous N states that is unresolved.  The first semantic option of a state
+ *  is always chosen.  */
+static void
+yyresolveLocations (yyGLRState* yys, int yyn,
+                   yyGLRStack *yystackp]b4_user_formals[)
+{
+  if (0 < yyn)
+    {
+      yyresolveLocations (yys->yypred, yyn-1, yystackp]b4_user_args[);
+      if (!yys->yyresolved)
+       {
+         yySemanticOption *yyoption;
+         yyGLRStackItem yyrhs[1 + YYMAXRHS];
+         int yynrhs;
+         int yychar_current;
+         YYSTYPE yylval_current;
+         YYLTYPE yylloc_current;
+         yyoption = yys->yysemantics.yyfirstVal;
+         YYASSERT (yyoption != NULL);
+         yynrhs = yyrhsLength (yyoption->yyrule);
+         if (yynrhs > 0)
+           {
+             yyGLRState *yys;
+             int yyi;
+             yyresolveLocations (yyoption->yystate, yynrhs,
+                                 yystackp]b4_user_args[);
+             for (yys = yyoption->yystate, yyi = yynrhs;
+                  yyi >= 1;
+                  yys = yys->yypred, yyi -= 1)
+               yyrhs[yyi].yystate.yyloc = yys->yyloc;
+           }
+         else
+           {
+             yyGLRState *yyprevious = yyoption->yystate;
+             YYASSERT (yyprevious->yyresolved);
+             yyrhs[0].yystate.yyloc = yyprevious->yyloc;
+           }
+         yychar_current = yychar;
+         yylval_current = yylval;
+         yylloc_current = yylloc;
+         yychar = yyoption->yyrawchar;
+         yylval = yyoption->yyval;
+         yylloc = yyoption->yyloc;
+         YYLLOC_DEFAULT ((yys->yyloc), yyrhs, yynrhs);
+         yychar = yychar_current;
+         yylval = yylval_current;
+         yylloc = yylloc_current;
+       }
+    }
+}
 
 /** Resolve the ambiguity represented in state S, perform the indicated
  *  actions, and set the semantic value of S.  If result != yyok, the chain of
@@ -1837,6 +1888,7 @@ yyresolveValue (yyGLRState* yys, yyGLRSt
          switch (yypreference (yybest, yyp))
            {
            case 0:
+             yyresolveLocations (yys, 1, yystackp]b4_user_args[);
              return yyreportAmbiguity (yybest, yyp]b4_pure_args[);
              break;
            case 1:
Index: doc/bison.texinfo
===================================================================
RCS file: /sources/bison/bison/doc/bison.texinfo,v
retrieving revision 1.177
diff -p -u -r1.177 bison.texinfo
--- doc/bison.texinfo   21 Feb 2006 21:01:57 -0000      1.177
+++ doc/bison.texinfo   6 Mar 2006 07:35:20 -0000
@@ -1135,13 +1135,16 @@ memory referenced by @code{yylval}.
 @findex YYERROR
 @cindex @acronym{GLR} parsers and @code{YYERROR}
 Another Bison feature requiring special consideration is @code{YYERROR}
-(@pxref{Action Features}), which you can invoke in any semantic action to
+(@pxref{Action Features}), which you can invoke in a semantic action to
 initiate error recovery.
 During deterministic @acronym{GLR} operation, the effect of @code{YYERROR} is
 the same as its effect in an @acronym{LALR}(1) parser.
 In a deferred semantic action, its effect is undefined.
 @c The effect is probably a syntax error at the split point.
 
+Also, see @ref{Location Default Action, ,Default Action for Locations}, which
+describes a special usage of @code{YYLLOC_DEFAULT} in @acronym{GLR} parsers.
+
 @node Compiler Requirements
 @subsection Considerations when Compiling @acronym{GLR} Parsers
 @cindex @code{inline}
@@ -3571,6 +3574,7 @@ This location is stored in @code{yylloc}
 @node Location Default Action
 @subsection Default Action for Locations
 @vindex YYLLOC_DEFAULT
address@hidden @acronym{GLR} parsers and @code{YYLLOC_DEFAULT}
 
 Actually, actions are not the best place to compute locations.  Since
 locations are much more general than semantic values, there is room in
@@ -3578,6 +3582,9 @@ the output parser to redefine the defaul
 rule.  The @code{YYLLOC_DEFAULT} macro is invoked each time a rule is
 matched, before the associated action is run.  It is also invoked
 while processing a syntax error, to compute the error's location.
+Before reporting an unresolvable syntactic ambiguity, a @acronym{GLR}
+parser invokes @code{YYLLOC_DEFAULT} recursively to compute the location
+of that ambiguity.
 
 Most of the time, this macro is general enough to suppress location
 dedicated code from semantic actions.
@@ -3586,9 +3593,11 @@ The @code{YYLLOC_DEFAULT} macro takes th
 the location of the grouping (the result of the computation).  When a
 rule is matched, the second parameter identifies locations of
 all right hand side elements of the rule being matched, and the third
-parameter is the size of the rule's right hand side.  When processing
-a syntax error, the second parameter identifies locations of
-the symbols that were discarded during error processing, and the third
+parameter is the size of the rule's right hand side.
+When a @acronym{GLR} parser reports an ambiguity, which of multiple candidate
+right hand sides it passes to @code{YYLLOC_DEFAULT} is undefined.
+When processing a syntax error, the second parameter identifies locations
+of the symbols that were discarded during error processing, and the third
 parameter is the number of discarded symbols.
 
 By default, @code{YYLLOC_DEFAULT} is defined this way:
Index: tests/glr-regression.at
===================================================================
RCS file: /sources/bison/bison/tests/glr-regression.at,v
retrieving revision 1.32
diff -p -u -r1.32 glr-regression.at
--- tests/glr-regression.at     4 Mar 2006 03:29:03 -0000       1.32
+++ tests/glr-regression.at     6 Mar 2006 07:35:20 -0000
@@ -1576,3 +1576,82 @@ AT_CHECK([[./glr-regr16]], 0, [],
 ])
 
 AT_CLEANUP
+
+
+## ------------------------------------------------------------------------- ##
+## Uninitialized location when reporting ambiguity.                         ##
+## ------------------------------------------------------------------------- ##
+
+AT_SETUP([Uninitialized location when reporting ambiguity])
+AT_DATA_GRAMMAR([glr-regr17.y],
+[[
+%glr-parser
+%locations
+%pure-parser
+%error-verbose
+
+%union { int dummy; }
+
+%{
+  static void yyerror (YYLTYPE *, char const *);
+  static int yylex (YYSTYPE *, YYLTYPE *);
+%}
+
+%initial-action {
+  @$.first_line = 1;
+  @$.first_column = 1;
+  @$.last_line = 1;
+  @$.last_column = 1;
+}
+
+%%
+
+/* Tests multiple levels of yyresolveLocations recursion.  */
+start: ambig1 | ambig2 ;
+ambig1: sub_ambig1 | sub_ambig2 ;
+ambig2: sub_ambig1 | sub_ambig2 ;
+
+/* Tests non-empty RHS as well as empty RHS with either initial location or
+   location of previous token.  Both empty RHS locations are printed in the
+   error message.  */
+sub_ambig1: empty 'a' 'b' empty ;
+sub_ambig2: empty 'a' 'b' empty ;
+empty: ;
+
+%%
+
+static void
+yyerror (YYLTYPE *locp, char const *msg)
+{
+  fprintf (stderr, "Error at %d.%d-%d.%d: %s.\n", locp->first_line,
+          locp->first_column, locp->last_line, locp->last_column, msg);
+}
+
+/*ARGSUSED*/ static int
+yylex (YYSTYPE *lvalp, YYLTYPE *llocp)
+{
+  static const char input[] = "ab";
+  static const char *inputp = input;
+  llocp->first_line = llocp->last_line = 2;
+  llocp->first_column = inputp - input + 1;
+  llocp->last_column = llocp->first_column + 1;
+  return *inputp++;
+}
+
+int
+main (void)
+{
+  return yyparse () != 1;
+}
+]])
+
+AT_CHECK([[bison -o glr-regr17.c glr-regr17.y]], 0, [],
+[glr-regr17.y: conflicts: 3 reduce/reduce
+])
+AT_COMPILE([glr-regr17])
+
+AT_CHECK([[./glr-regr17]], 0, [],
+[Error at 1.1-2.3: syntax is ambiguous.
+])
+
+AT_CLEANUP