mirror of
https://sourceware.org/git/binutils-gdb.git
synced 2024-12-25 18:23:29 +08:00
356 lines
8.4 KiB
Plaintext
356 lines
8.4 KiB
Plaintext
%{ /* mcparse.y -- parser for Windows mc files
|
|
Copyright (C) 2007-2018 Free Software Foundation, Inc.
|
|
|
|
Parser for Windows mc files
|
|
Written by Kai Tietz, Onevision.
|
|
|
|
This file is part of GNU Binutils.
|
|
|
|
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 3 of the License, or
|
|
(at your option) any later version.
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
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. */
|
|
|
|
/* This is a parser for Windows rc files. It is based on the parser
|
|
by Gunther Ebert <gunther.ebert@ixos-leipzig.de>. */
|
|
|
|
#include "sysdep.h"
|
|
#include "bfd.h"
|
|
#include "bucomm.h"
|
|
#include "libiberty.h"
|
|
#include "windmc.h"
|
|
#include "safe-ctype.h"
|
|
|
|
static rc_uint_type mc_last_id = 0;
|
|
static rc_uint_type mc_sefa_val = 0;
|
|
static unichar *mc_last_symbol = NULL;
|
|
static const mc_keyword *mc_cur_severity = NULL;
|
|
static const mc_keyword *mc_cur_facility = NULL;
|
|
static mc_node *cur_node = NULL;
|
|
|
|
%}
|
|
|
|
%union
|
|
{
|
|
rc_uint_type ival;
|
|
unichar *ustr;
|
|
const mc_keyword *tok;
|
|
mc_node *nod;
|
|
};
|
|
|
|
%start input
|
|
|
|
%token NL
|
|
%token<ustr> MCIDENT MCFILENAME MCLINE MCCOMMENT
|
|
%token<tok> MCTOKEN
|
|
%token MCENDLINE
|
|
%token MCLANGUAGENAMES MCFACILITYNAMES MCSEVERITYNAMES MCOUTPUTBASE MCMESSAGEIDTYPEDEF
|
|
%token MCLANGUAGE MCMESSAGEID MCSEVERITY MCFACILITY MCSYMBOLICNAME
|
|
%token <ival> MCNUMBER
|
|
|
|
%type<ival> id vid sefasy_def
|
|
%type<ustr> alias_name token lines comments
|
|
%type<tok> lang
|
|
|
|
%%
|
|
input: entities
|
|
;
|
|
|
|
entities:
|
|
/* empty */
|
|
| entities entity
|
|
;
|
|
entity: global_section
|
|
| message
|
|
| comments
|
|
{
|
|
cur_node = mc_add_node ();
|
|
cur_node->user_text = $1;
|
|
}
|
|
| error { mc_fatal ("syntax error"); }
|
|
;
|
|
|
|
global_section:
|
|
MCSEVERITYNAMES '=' '(' severitymaps ')'
|
|
| MCSEVERITYNAMES '=' '(' severitymaps error { mc_fatal ("missing ')' in SeverityNames"); }
|
|
| MCSEVERITYNAMES '=' error { mc_fatal ("missing '(' in SeverityNames"); }
|
|
| MCSEVERITYNAMES error { mc_fatal ("missing '=' for SeverityNames"); }
|
|
| MCLANGUAGENAMES '=' '(' langmaps ')'
|
|
| MCLANGUAGENAMES '=' '(' langmaps error { mc_fatal ("missing ')' in LanguageNames"); }
|
|
| MCLANGUAGENAMES '=' error { mc_fatal ("missing '(' in LanguageNames"); }
|
|
| MCLANGUAGENAMES error { mc_fatal ("missing '=' for LanguageNames"); }
|
|
| MCFACILITYNAMES '=' '(' facilitymaps ')'
|
|
| MCFACILITYNAMES '=' '(' facilitymaps error { mc_fatal ("missing ')' in FacilityNames"); }
|
|
| MCFACILITYNAMES '=' error { mc_fatal ("missing '(' in FacilityNames"); }
|
|
| MCFACILITYNAMES error { mc_fatal ("missing '=' for FacilityNames"); }
|
|
| MCOUTPUTBASE '=' MCNUMBER
|
|
{
|
|
if ($3 != 10 && $3 != 16)
|
|
mc_fatal ("OutputBase allows 10 or 16 as value");
|
|
mcset_out_values_are_decimal = ($3 == 10 ? 1 : 0);
|
|
}
|
|
| MCMESSAGEIDTYPEDEF '=' MCIDENT
|
|
{
|
|
mcset_msg_id_typedef = $3;
|
|
}
|
|
| MCMESSAGEIDTYPEDEF '=' error
|
|
{
|
|
mc_fatal ("MessageIdTypedef expects an identifier");
|
|
}
|
|
| MCMESSAGEIDTYPEDEF error
|
|
{
|
|
mc_fatal ("missing '=' for MessageIdTypedef");
|
|
}
|
|
;
|
|
|
|
severitymaps:
|
|
severitymap
|
|
| severitymaps severitymap
|
|
| error { mc_fatal ("severity ident missing"); }
|
|
;
|
|
|
|
severitymap:
|
|
token '=' MCNUMBER alias_name
|
|
{
|
|
mc_add_keyword ($1, MCTOKEN, "severity", $3, $4);
|
|
}
|
|
| token '=' error { mc_fatal ("severity number missing"); }
|
|
| token error { mc_fatal ("severity missing '='"); }
|
|
;
|
|
|
|
facilitymaps:
|
|
facilitymap
|
|
| facilitymaps facilitymap
|
|
| error { mc_fatal ("missing ident in FacilityNames"); }
|
|
;
|
|
|
|
facilitymap:
|
|
token '=' MCNUMBER alias_name
|
|
{
|
|
mc_add_keyword ($1, MCTOKEN, "facility", $3, $4);
|
|
}
|
|
| token '=' error { mc_fatal ("facility number missing"); }
|
|
| token error { mc_fatal ("facility missing '='"); }
|
|
;
|
|
|
|
langmaps:
|
|
langmap
|
|
| langmaps langmap
|
|
| error { mc_fatal ("missing ident in LanguageNames"); }
|
|
;
|
|
|
|
langmap:
|
|
token '=' MCNUMBER lex_want_filename ':' MCFILENAME
|
|
{
|
|
mc_add_keyword ($1, MCTOKEN, "language", $3, $6);
|
|
}
|
|
| token '=' MCNUMBER lex_want_filename ':' error { mc_fatal ("missing filename in LanguageNames"); }
|
|
| token '=' MCNUMBER error { mc_fatal ("missing ':' in LanguageNames"); }
|
|
| token '=' error { mc_fatal ("missing language code in LanguageNames"); }
|
|
| token error { mc_fatal ("missing '=' for LanguageNames"); }
|
|
;
|
|
|
|
alias_name:
|
|
/* empty */
|
|
{
|
|
$$ = NULL;
|
|
}
|
|
| ':' MCIDENT
|
|
{
|
|
$$ = $2;
|
|
}
|
|
| ':' error { mc_fatal ("illegal token in identifier"); $$ = NULL; }
|
|
;
|
|
|
|
message:
|
|
id sefasy_def
|
|
{
|
|
cur_node = mc_add_node ();
|
|
cur_node->symbol = mc_last_symbol;
|
|
cur_node->facility = mc_cur_facility;
|
|
cur_node->severity = mc_cur_severity;
|
|
cur_node->id = ($1 & 0xffffUL);
|
|
cur_node->vid = ($1 & 0xffffUL) | mc_sefa_val;
|
|
mc_last_id = $1;
|
|
}
|
|
lang_entities
|
|
;
|
|
|
|
id: MCMESSAGEID '=' vid { $$ = $3; }
|
|
| MCMESSAGEID '=' error { mc_fatal ("missing number in MessageId"); $$ = 0; }
|
|
| MCMESSAGEID error { mc_fatal ("missing '=' for MessageId"); $$ = 0; }
|
|
;
|
|
|
|
vid: /* empty */
|
|
{
|
|
$$ = ++mc_last_id;
|
|
}
|
|
| MCNUMBER
|
|
{
|
|
$$ = $1;
|
|
}
|
|
| '+' MCNUMBER
|
|
{
|
|
$$ = mc_last_id + $2;
|
|
}
|
|
| '+' error { mc_fatal ("missing number after MessageId '+'"); }
|
|
;
|
|
|
|
sefasy_def:
|
|
/* empty */
|
|
{
|
|
$$ = 0;
|
|
mc_sefa_val = (mcset_custom_bit ? 1 : 0) << 29;
|
|
mc_last_symbol = NULL;
|
|
mc_cur_severity = NULL;
|
|
mc_cur_facility = NULL;
|
|
}
|
|
| sefasy_def severity
|
|
{
|
|
if ($1 & 1)
|
|
mc_warn (_("duplicate definition of Severity"));
|
|
$$ = $1 | 1;
|
|
}
|
|
| sefasy_def facility
|
|
{
|
|
if ($1 & 2)
|
|
mc_warn (_("duplicate definition of Facility"));
|
|
$$ = $1 | 2;
|
|
}
|
|
| sefasy_def symbol
|
|
{
|
|
if ($1 & 4)
|
|
mc_warn (_("duplicate definition of SymbolicName"));
|
|
$$ = $1 | 4;
|
|
}
|
|
;
|
|
|
|
severity: MCSEVERITY '=' MCTOKEN
|
|
{
|
|
mc_sefa_val &= ~ (0x3UL << 30);
|
|
mc_sefa_val |= (($3->nval & 0x3UL) << 30);
|
|
mc_cur_severity = $3;
|
|
}
|
|
;
|
|
|
|
facility: MCFACILITY '=' MCTOKEN
|
|
{
|
|
mc_sefa_val &= ~ (0xfffUL << 16);
|
|
mc_sefa_val |= (($3->nval & 0xfffUL) << 16);
|
|
mc_cur_facility = $3;
|
|
}
|
|
;
|
|
|
|
symbol: MCSYMBOLICNAME '=' MCIDENT
|
|
{
|
|
mc_last_symbol = $3;
|
|
}
|
|
;
|
|
|
|
lang_entities:
|
|
lang_entity
|
|
| lang_entities lang_entity
|
|
;
|
|
|
|
lang_entity:
|
|
lang lex_want_line lines MCENDLINE
|
|
{
|
|
mc_node_lang *h;
|
|
h = mc_add_node_lang (cur_node, $1, cur_node->vid);
|
|
h->message = $3;
|
|
if (mcset_max_message_length != 0 && unichar_len (h->message) > mcset_max_message_length)
|
|
mc_warn ("message length to long");
|
|
}
|
|
;
|
|
|
|
lines: MCLINE
|
|
{
|
|
$$ = $1;
|
|
}
|
|
| lines MCLINE
|
|
{
|
|
unichar *h;
|
|
rc_uint_type l1,l2;
|
|
l1 = unichar_len ($1);
|
|
l2 = unichar_len ($2);
|
|
h = (unichar *) res_alloc ((l1 + l2 + 1) * sizeof (unichar));
|
|
if (l1) memcpy (h, $1, l1 * sizeof (unichar));
|
|
if (l2) memcpy (&h[l1], $2, l2 * sizeof (unichar));
|
|
h[l1 + l2] = 0;
|
|
$$ = h;
|
|
}
|
|
| error { mc_fatal ("missing end of message text"); $$ = NULL; }
|
|
| lines error { mc_fatal ("missing end of message text"); $$ = $1; }
|
|
;
|
|
|
|
comments: MCCOMMENT { $$ = $1; }
|
|
| comments MCCOMMENT
|
|
{
|
|
unichar *h;
|
|
rc_uint_type l1,l2;
|
|
l1 = unichar_len ($1);
|
|
l2 = unichar_len ($2);
|
|
h = (unichar *) res_alloc ((l1 + l2 + 1) * sizeof (unichar));
|
|
if (l1) memcpy (h, $1, l1 * sizeof (unichar));
|
|
if (l2) memcpy (&h[l1], $2, l2 * sizeof (unichar));
|
|
h[l1 + l2] = 0;
|
|
$$ = h;
|
|
}
|
|
;
|
|
|
|
lang: MCLANGUAGE lex_want_nl '=' MCTOKEN NL
|
|
{
|
|
$$ = $4;
|
|
}
|
|
| MCLANGUAGE lex_want_nl '=' MCIDENT NL
|
|
{
|
|
$$ = NULL;
|
|
mc_fatal (_("undeclared language identifier"));
|
|
}
|
|
| MCLANGUAGE lex_want_nl '=' token error
|
|
{
|
|
$$ = NULL;
|
|
mc_fatal ("missing newline after Language");
|
|
}
|
|
| MCLANGUAGE lex_want_nl '=' error
|
|
{
|
|
$$ = NULL;
|
|
mc_fatal ("missing ident for Language");
|
|
}
|
|
| MCLANGUAGE error
|
|
{
|
|
$$ = NULL;
|
|
mc_fatal ("missing '=' for Language");
|
|
}
|
|
;
|
|
|
|
token: MCIDENT { $$ = $1; }
|
|
| MCTOKEN { $$ = $1->usz; }
|
|
;
|
|
|
|
lex_want_nl:
|
|
/* Empty */ { mclex_want_nl = 1; }
|
|
;
|
|
|
|
lex_want_line:
|
|
/* Empty */ { mclex_want_line = 1; }
|
|
;
|
|
|
|
lex_want_filename:
|
|
/* Empty */ { mclex_want_filename = 1; }
|
|
;
|
|
|
|
%%
|
|
|
|
/* Something else. */
|