]> git.saurik.com Git - bison.git/blobdiff - src/scan-gram.l
Factor %FLAG at scan level.
[bison.git] / src / scan-gram.l
index 6fa9f220f3ec5cc9f09f967ca768caf0a4379b30..2cfbb4e4c1b96bb3fd6fd71d3082e436a7b70f2a 100644 (file)
@@ -1,12 +1,13 @@
 /* Bison Grammar Scanner                             -*- C -*-
 
-   Copyright (C) 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
+   Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
+   Free Software Foundation, Inc.
 
    This file is part of Bison, the GNU Compiler Compiler.
 
-   This program is free software; you can redistribute it and/or modify
+   This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
-   the Free Software Foundation; either version 2 of the License, or
+   the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.
 
    This program is distributed in the hope that it will be useful,
    GNU General Public License for more details.
 
    You should have received a copy of the GNU General Public License
-   along with this program; if not, write to the Free Software
-   Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
-   02110-1301  USA
-*/
+   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
 
 %option debug nodefault nounput noyywrap never-interactive
 %option prefix="gram_" outfile="lex.yy.c"
 #define gram_wrap() 1
 
 #define FLEX_PREFIX(Id) gram_ ## Id
-#include "flex-scanner.h"
+#include <src/flex-scanner.h>
 
-#include "complain.h"
-#include "files.h"
-#include "gram.h"
-#include "quotearg.h"
-#include "reader.h"
-#include "uniqstr.h"
+#include <src/complain.h>
+#include <src/files.h>
+#include <src/gram.h>
+#include <quotearg.h>
+#include <src/reader.h>
+#include <src/uniqstr.h>
 
 #include <mbswidth.h>
 #include <quote.h>
 
-#include "scan-gram.h"
+#include <src/scan-gram.h>
 
 #define YY_DECL GRAM_LEX_DECL
 
@@ -57,8 +55,15 @@ static boundary scanner_cursor;
 static size_t no_cr_read (FILE *, char *, size_t);
 #define YY_INPUT(buf, result, size) ((result) = no_cr_read (yyin, buf, size))
 
+#define RETURN_PERCENT_FLAG(Value)              \
+  do {                                          \
+    val->uniqstr = uniqstr_new (Value);         \
+    return PERCENT_FLAG;                        \
+  } while (0)
+
+
 /* A string representing the most recently saved token.  */
-char *last_string;
+static char *last_string;
 
 void
 gram_scanner_last_string_free (void)
@@ -80,6 +85,8 @@ static void unexpected_newline (boundary, char const *);
  /* A identifier was just read in directives/rules.  Special state
     to capture the sequence `identifier :'. */
 %x SC_AFTER_IDENTIFIER
+ /* A complex tag, with nested angles brackets. */
+%x SC_TAG
 
  /* Three types of user code:
     - prologue (code between `%{' `%}' in the first section, before %%);
@@ -98,8 +105,10 @@ int   [0-9]+
 
 /* POSIX says that a tag must be both an id and a C union member, but
    historically almost any character is allowed in a tag.  We disallow
-   NUL and newline, as this simplifies our implementation.  */
-tag     [^\0\n>]+
+   NUL, as this simplifies our implementation.  We disallow angle
+   bracket to match them in nested pairs: several languages use them
+   for generics/template types.  */
+tag     [^\0<>]+
 
 /* Zero or more instances of backslash-newline.  Following GCC, allow
    white space between the backslash and the newline.  */
@@ -107,8 +116,9 @@ splice       (\\[ \f\t\v]*\n)*
 
 %%
 %{
-  /* Nesting level of the current code in braces.  */
-  int braces_level IF_LINT (= 0);
+  /* Nesting level.  Either for nested braces, or nested angle brackets
+     (but not mixed).  */
+  int nesting IF_LINT (= 0);
 
   /* Parent context state, when applicable.  */
   int context_state IF_LINT (= 0);
@@ -154,53 +164,58 @@ splice     (\\[ \f\t\v]*\n)*
   /*----------------------------.
   | Scanning Bison directives.  |
   `----------------------------*/
+
+  /* For directives that are also command line options, the regex must be
+       "%..."
+     after "[-_]"s are removed, and the directive must match the --long
+     option name, with a single string argument.  Otherwise, add exceptions
+     to ../build-aux/cross-options.pl.  */
+
 <INITIAL>
 {
-  "%binary"                        return PERCENT_NONASSOC;
+  "%binary"                         return PERCENT_NONASSOC;
   "%code"                           return PERCENT_CODE;
-  "%code-top"                       return PERCENT_CODE_TOP;
-  "%debug"                         return PERCENT_DEBUG;
-  "%default"[-_]"prec"             return PERCENT_DEFAULT_PREC;
-  "%define"                        return PERCENT_DEFINE;
-  "%defines"                       return PERCENT_DEFINES;
-  "%destructor"                            return PERCENT_DESTRUCTOR;
-  "%dprec"                         return PERCENT_DPREC;
-  "%error"[-_]"verbose"            return PERCENT_ERROR_VERBOSE;
-  "%expect"                        return PERCENT_EXPECT;
-  "%expect"[-_]"rr"                return PERCENT_EXPECT_RR;
-  "%file-prefix"                   return PERCENT_FILE_PREFIX;
+  "%debug"                          RETURN_PERCENT_FLAG("debug");
+  "%default"[-_]"prec"              return PERCENT_DEFAULT_PREC;
+  "%define"                         return PERCENT_DEFINE;
+  "%defines"                        return PERCENT_DEFINES;
+  "%destructor"                     return PERCENT_DESTRUCTOR;
+  "%dprec"                          return PERCENT_DPREC;
+  "%error"[-_]"verbose"             RETURN_PERCENT_FLAG("error_verbose");
+  "%expect"                         return PERCENT_EXPECT;
+  "%expect"[-_]"rr"                 return PERCENT_EXPECT_RR;
+  "%file-prefix"                    return PERCENT_FILE_PREFIX;
   "%fixed"[-_]"output"[-_]"files"   return PERCENT_YACC;
-  "%initial-action"                return PERCENT_INITIAL_ACTION;
-  "%glr-parser"                    return PERCENT_GLR_PARSER;
-  "%left"                          return PERCENT_LEFT;
-  "%lex-param"                     return PERCENT_LEX_PARAM;
-  "%locations"                     return PERCENT_LOCATIONS;
-  "%merge"                         return PERCENT_MERGE;
-  "%name"[-_]"prefix"              return PERCENT_NAME_PREFIX;
-  "%no"[-_]"default"[-_]"prec"     return PERCENT_NO_DEFAULT_PREC;
-  "%no"[-_]"lines"                 return PERCENT_NO_LINES;
-  "%nonassoc"                      return PERCENT_NONASSOC;
-  "%nondeterministic-parser"               return PERCENT_NONDETERMINISTIC_PARSER;
-  "%nterm"                         return PERCENT_NTERM;
-  "%output"                        return PERCENT_OUTPUT;
-  "%parse-param"                   return PERCENT_PARSE_PARAM;
-  "%prec"                          return PERCENT_PREC;
-  "%printer"                       return PERCENT_PRINTER;
-  "%provides"                       return PERCENT_PROVIDES;
-  "%pure"[-_]"parser"              return PERCENT_PURE_PARSER;
-  "%push"[-_]"parser"              return PERCENT_PUSH_PARSER;
-  "%require"                       return PERCENT_REQUIRE;
-  "%requires"                       return PERCENT_REQUIRES;
-  "%right"                         return PERCENT_RIGHT;
-  "%skeleton"                      return PERCENT_SKELETON;
-  "%start"                         return PERCENT_START;
-  "%term"                          return PERCENT_TOKEN;
-  "%token"                         return PERCENT_TOKEN;
-  "%token"[-_]"table"              return PERCENT_TOKEN_TABLE;
-  "%type"                          return PERCENT_TYPE;
-  "%union"                         return PERCENT_UNION;
-  "%verbose"                       return PERCENT_VERBOSE;
-  "%yacc"                          return PERCENT_YACC;
+  "%initial-action"                 return PERCENT_INITIAL_ACTION;
+  "%glr-parser"                     return PERCENT_GLR_PARSER;
+  "%language"                       return PERCENT_LANGUAGE;
+  "%left"                           return PERCENT_LEFT;
+  "%lex-param"                      return PERCENT_LEX_PARAM;
+  "%locations"                      return PERCENT_LOCATIONS;
+  "%merge"                          return PERCENT_MERGE;
+  "%name"[-_]"prefix"               return PERCENT_NAME_PREFIX;
+  "%no"[-_]"default"[-_]"prec"      return PERCENT_NO_DEFAULT_PREC;
+  "%no"[-_]"lines"                  return PERCENT_NO_LINES;
+  "%nonassoc"                       return PERCENT_NONASSOC;
+  "%nondeterministic-parser"        return PERCENT_NONDETERMINISTIC_PARSER;
+  "%nterm"                          return PERCENT_NTERM;
+  "%output"                         return PERCENT_OUTPUT;
+  "%parse-param"                    return PERCENT_PARSE_PARAM;
+  "%prec"                           return PERCENT_PREC;
+  "%precedence"                     return PERCENT_PRECEDENCE;
+  "%printer"                        return PERCENT_PRINTER;
+  "%pure"[-_]"parser"               return PERCENT_PURE_PARSER;
+  "%require"                        return PERCENT_REQUIRE;
+  "%right"                          return PERCENT_RIGHT;
+  "%skeleton"                       return PERCENT_SKELETON;
+  "%start"                          return PERCENT_START;
+  "%term"                           return PERCENT_TOKEN;
+  "%token"                          return PERCENT_TOKEN;
+  "%token"[-_]"table"               return PERCENT_TOKEN_TABLE;
+  "%type"                           return PERCENT_TYPE;
+  "%union"                          return PERCENT_UNION;
+  "%verbose"                        return PERCENT_VERBOSE;
+  "%yacc"                           return PERCENT_YACC;
 
   {directive} {
     complain_at (*loc, _("invalid directive: %s"), quote (yytext));
@@ -209,8 +224,6 @@ splice       (\\[ \f\t\v]*\n)*
   "="                     return EQUAL;
   "|"                     return PIPE;
   ";"                     return SEMICOLON;
-  "<*>"                   return TYPE_TAG_ANY;
-  "<!>"                   return TYPE_TAG_NONE;
 
   {id} {
     val->uniqstr = uniqstr_new (yytext);
@@ -239,18 +252,25 @@ splice     (\\[ \f\t\v]*\n)*
   /* Code in between braces.  */
   "{" {
     STRING_GROW;
-    braces_level = 0;
+    nesting = 0;
     code_start = loc->start;
     BEGIN SC_BRACED_CODE;
   }
 
   /* A type. */
+  "<*>"       return TAG_ANY;
+  "<>"        return TAG_NONE;
   "<"{tag}">" {
     obstack_grow (&obstack_for_string, yytext + 1, yyleng - 2);
     STRING_FINISH;
     val->uniqstr = uniqstr_new (last_string);
     STRING_FREE;
-    return TYPE;
+    return TAG;
+  }
+  "<"         {
+    nesting = 0;
+    token_start = loc->start;
+    BEGIN SC_TAG;
   }
 
   "%%" {
@@ -271,6 +291,17 @@ splice      (\\[ \f\t\v]*\n)*
 }
 
 
+  /*--------------------------------------------------------------.
+  | Supporting \0 complexifies our implementation for no expected |
+  | added value.                                                  |
+  `--------------------------------------------------------------*/
+
+<SC_ESCAPED_CHARACTER,SC_ESCAPED_STRING,SC_TAG>
+{
+  \0       complain_at (*loc, _("invalid null character"));
+}
+
+
   /*-----------------------------------------------------------------.
   | Scanning after an identifier, checking whether a colon is next.  |
   `-----------------------------------------------------------------*/
@@ -390,11 +421,40 @@ splice     (\\[ \f\t\v]*\n)*
   }
 }
 
-<SC_ESCAPED_CHARACTER,SC_ESCAPED_STRING>
+  /*-----------------------------------------------------------.
+  | Scanning a Bison nested tag.  The initial angle bracket is |
+  | already eaten.                                             |
+  `-----------------------------------------------------------*/
+
+<SC_TAG>
 {
-  \0       complain_at (*loc, _("invalid null character"));
-}
+  ">" {
+    --nesting;
+    if (nesting < 0)
+      {
+        STRING_FINISH;
+        loc->start = token_start;
+        val->uniqstr = uniqstr_new (last_string);
+        STRING_FREE;
+        BEGIN INITIAL;
+        return TAG;
+      }
+    STRING_GROW;
+  }
+
+  [^<>]+ STRING_GROW;
+  "<"+   STRING_GROW; nesting += yyleng;
 
+  <<EOF>> {
+    unexpected_eof (token_start, ">");
+    STRING_FINISH;
+    loc->start = token_start;
+    val->uniqstr = uniqstr_new (last_string);
+    STRING_FREE;
+    BEGIN INITIAL;
+    return TAG;
+  }
+}
 
   /*----------------------------.
   | Decode escaped characters.  |
@@ -513,13 +573,13 @@ splice     (\\[ \f\t\v]*\n)*
 
 <SC_BRACED_CODE>
 {
-  "{"|"<"{splice}"%"  STRING_GROW; braces_level++;
-  "%"{splice}">"      STRING_GROW; braces_level--;
+  "{"|"<"{splice}"%"  STRING_GROW; nesting++;
+  "%"{splice}">"      STRING_GROW; nesting--;
   "}" {
     obstack_1grow (&obstack_for_string, '}');
 
-    --braces_level;
-    if (braces_level < 0)
+    --nesting;
+    if (nesting < 0)
       {
        STRING_FINISH;
        loc->start = code_start;
@@ -800,5 +860,5 @@ gram_scanner_free (void)
 {
   obstack_free (&obstack_for_string, 0);
   /* Reclaim Flex's buffers.  */
-  yy_delete_buffer (YY_CURRENT_BUFFER);
+  yylex_destroy ();
 }