Author: kjs
Date: Mon Oct 1 10:50:41 2007
New Revision: 21719
Modified:
trunk/languages/PIR/lib/pir.pg
Log:
languages/PIR:
* refactor rules/tokens to handle syntax_errors only once
* cleanup grammar.
Modified: trunk/languages/PIR/lib/pir.pg
==============================================================================
--- trunk/languages/PIR/lib/pir.pg (original)
+++ trunk/languages/PIR/lib/pir.pg Mon Oct 1 10:50:41 2007
@@ -18,10 +18,10 @@
}
rule program {
- <?nl>*
+ <?NL>*
<compilation_unit>
- [ <?nl> <compilation_unit> ]*
- [ <?nl> | <warning: no newline at end of file> ]
+ [ <?NL> <compilation_unit> ]*
+ [ <?NL> | <warning: no newline at end of file> ]
}
token compilation_unit {
@@ -41,15 +41,15 @@
<'.sub'>
<sub_id>
<sub_pragmas>?
- <?NL>
+ <?nl>
<param_decl>*
<labeled_pir_instr>*
[ <'.end'> | <syntax_error: '.end' expected to close method/sub body> ]
}
rule sub_id {
- | <id>
- | <string_constant>
+ | <IDENT>
+ | <STRING_CONST>
| <syntax_error: subroutine identifier expected>
}
@@ -80,22 +80,30 @@
rule parenthesized_string {
<'('>
- [ <string_constant> | <syntax_error: string constant expected> ]
- [ <')'> | <syntax_error: ')' expected> ]
+ <string_constant>
+ <?right_paren>
}
rule multi_pragma {
<':multi'>
- [ <'('> | <syntax_error: '(' expected after ':multi' flag> ]
+ <?left_paren>
<multi_types>?
- [ <')'> | <syntax_error: ')' expected to close ':multi' flag> ]
+ <?right_paren>
+}
+
+rule left_paren {
+ <'('> | <syntax_error: '(' expected>
+}
+
+rule right_paren {
+ <')'> | <syntax_error: ')' expected>
}
rule outer_pragma {
<':outer'>
- [ <'('> | <syntax_error: '(' expected after ':outer' flag> ]
+ <?left_paren>
<sub_id>
- [ <')'> | <syntax_error: ')' expected to close ':outer' flag> ]
+ <?right_paren>
}
rule multi_types {
@@ -106,18 +114,18 @@
| <type>
| <'_'>
| <keylist>
- | <id>
- | <string_constant>
+ | <IDENT>
+ | <STRING_CONST>
}
rule param_decl {
<'.param'>
[ [ <type> <id> ]
- | <reg>
+ | <REGISTER>
| <syntax_error: parameter type or register expected>
]
[ <get_flags> | <':unique_reg'> ]*
- <?NL>
+ <?nl>
}
@@ -132,14 +140,14 @@
[ <label> <instr>?
| <instr>
]
- <?NL>
+ <?nl>
}
rule labeled_pasm_instr {
[ <label> <pasm_instr>?
| <pasm_instr>
]
- <?NL>
+ <?nl>
}
rule instr {
@@ -195,8 +203,8 @@
rule lexical_decl {
<'.lex'>
- <string_constant1>
- <comma>
+ <string_constant>
+ <?comma>
<target1>
}
@@ -204,9 +212,6 @@
<','> | <syntax_error: ',' expected>
}
-rule string_constant1 {
- <string_constant> | <syntax_error: string constant expected>
-}
rule target1 {
<target> | <syntax_error: register or identifier expected>
@@ -227,16 +232,20 @@
}
rule const_def_tail {
- | <'int'> <const_assign> [ <int_constant> | <syntax_error: integer
constant expected> ]
- | <'num'> <const_assign> [ <float_constant> | <syntax_error: floating
point constant expected> ]
- | <'pmc'> <const_assign> [ <string_constant> | <syntax_error: string
constant expected> ]
- | <'string'> <const_assign> [ <string_constant> | <syntax_error: string
constant expected> ]
+ | <'int'> <const_assign> <int_constant>
+ | <'num'> <const_assign> <float_constant>
+ | <'pmc'> <const_assign> <string_constant>
+ | <'string'> <const_assign> <string_constant>
| <syntax_error: type expected for constant>
}
rule const_assign {
- [ <id> | <syntax_error: constant identifier expected> ]
- [ <'='> | <syntax_error: '=' expected after constant identifier> ]
+ <id>
+ <?assign_token>
+}
+
+rule assign_token {
+ <'='> | <syntax_error: '=' expected>
}
rule conditional_stat {
@@ -248,7 +257,7 @@
}
token label_identifier {
- | <id>
+ | <IDENT>
| <pasm_instr>
}
@@ -294,7 +303,7 @@
rule keylist {
<'['>
<key> [ <separator> <key> ]*
- [ <']'> | <syntax_error: ']' expected> ]
+ <right_bracket>
}
token separator {
@@ -318,7 +327,7 @@
# #| <target> <'='> <pasm_op_2> <simple_expr> \, <simple_expr>
# | <target> <'='> <'new'> <string_constant>
# | <target> <'='> <'new'> <keylist>
-# | <target> <'='> <'find_type'> [ <string_constant> | <string_reg> | <id> ]
+# | <target> <'='> <'find_type'> [ <string_constant> | <string_reg> |
<IDENT> ]
# | <target> <'='> <heredoc>
# | <target> <assign_operator> <simple_expr>
# | <target> <keylist> <'='> <simple_expr>
@@ -341,8 +350,7 @@
| <expression>
| <pasm_instruction> \N*
| <heredoc>
- | <'new'> <string_constant>
- | <'new'> <keylist>
+ | <'new'> [ <STRING_CONST> | <keylist> ]
}
@@ -371,15 +379,15 @@
rule heredoc {
<'<<'>
- [ <string_constant> | <syntax_error: heredoc label identifier expected> ]
- [ <?nl> | <syntax_error: newline expected after heredoc label> ]
+ <string_constant>
+ <?nl>
<heredoc_string>
[ <heredoc_label> | <syntax_error: heredoc label expected> ]
}
token heredoc_label {
[ ^^ | <syntax_error: no whitespace allowed in front of a heredoc label> ]
- <id>
+ <IDENT>
[ $$ | <syntax_error: no whitespace allowed after a heredoc> ]
}
@@ -390,7 +398,7 @@
rule long_sub_call {
<'.pcc_begin'>
- [ <?nl> | <syntax_error: newline after '.pcc_begin' expected> ]
+ <?nl>
<arguments>
[
[ <'.pcc_call'>
@@ -398,8 +406,8 @@
]
| [ <invocant> <'.meth_call'> ]
]
- [ <target> | <syntax_error: id or register expected that holds the sub
object> ]
- [ <?nl> | <syntax_error: newline after '.pcc_call sub' expected> ]
+ [ <target> | <syntax_error: IDENT or register expected that holds the sub
object> ]
+ <?nl>
[ <local_decl> <?nl> ]*
<result_values>
[ <'.pcc_end'> | <syntax_error: '.pcc_end' expected> ]
@@ -423,7 +431,7 @@
rule short_sub_call {
[
[ <target>\.]? # optional invocant
- [ <target> | <string_constant> ] # method or sub name/id
+ [ <target> | <STRING_CONST> ] # method or sub name/IDENT
|
<target> <method>
]
@@ -434,11 +442,12 @@
token method {
| \.<target>
- | \.<string_constant>
+ | \.<STRING_CONST>
}
rule sub_invocation {
- <long_sub_call> | <short_sub_call>
+ | <long_sub_call>
+ | <short_sub_call>
}
rule result_var_list {
@@ -449,7 +458,7 @@
[ <result_var> | <syntax_error: target to store result expected after
','> ]
]*
]?
- [ <')'> | <syntax_error: ')' expected to close result target list> ]
+ <right_paren>
}
rule result_var {
@@ -464,13 +473,13 @@
[ <arg> | <syntax_error: argument expected after ','> ]
]*
]?
- [ <')'> | <syntax_error: ')' expected to close argument list> ]
+ <right_paren>
}
rule arg {
- [ <float_constant>
- | <int_constant>
- | <string_constant>
+ [ <FLOAT_CONST>
+ | <INT_CONST>
+ | <STRING_CONST>
[ <'=>'>
[ <target> | <syntax_error: target for named argument expected> ]
]?
@@ -491,7 +500,7 @@
[ <'.arg'>
[ <simple_expr> | <syntax_error: argument expression expected> ]
<set_flags>?
- [ <?nl> | <syntax_error: newline expected after .arg statement> ]
+ <?nl>
]*
}
@@ -499,7 +508,7 @@
[ <'.result'>
[ <target> | <syntax_error: target expected to hold the result> ]
<get_flags>?
- [ <?nl> | <syntax_error: newline expected after .result statement> ]
+ <?nl>
]*
}
@@ -516,7 +525,7 @@
}
rule named_flag {
- <':named'> <parenthesized_string>?
+ <':named'> <parenthesized_string>?
}
token return_stat {
@@ -528,7 +537,7 @@
}
rule long_return_stat {
- <'.pcc_begin_return'> <?NL>
+ <'.pcc_begin_return'> <?nl>
[ <'.return'> <return_value> ]*
[ <'.pcc_end_return'> | <syntax_error: '.pcc_end_return' expected> ]
}
@@ -538,7 +547,7 @@
}
rule long_yield_stat {
- <'.pcc_begin_yield'> <?NL>
+ <'.pcc_begin_yield'> <?nl>
[ <'.yield'> <return_value> ]*
[ <'.pcc_end_yield'> | <syntax_error: '.pcc_end_yield' expected> ]
}
@@ -547,7 +556,7 @@
rule return_value {
[ <simple_expr> | <syntax_error: simple expression expected> ]
<set_flags>?
- <?NL>
+ <?nl>
}
rule short_yield_stat {
@@ -559,36 +568,32 @@
}
rule open_namespace {
- <'.namespace'> <scope_id>
+ <'.namespace'> <id>
}
rule close_namespace {
- <'.endnamespace'> <scope_id>
-}
-
-rule scope_id {
- <id> | <syntax_error: namespace identifier expected>
+ <'.endnamespace'> <id>
}
rule emit {
<'.emit'>
- <?NL>
+ <?nl>
<labeled_pasm_instr>*
[ <'.eom'> | <syntax_error: '.eom' expected> ]
}
rule macro_def {
<'.macro'>
- [ <id> | <syntax_error: macro identifier expected> ]
+ <id>
<macro_parameters>?
- <?NL>
+ <?nl>
<macro_body>
}
rule macro_parameters {
<'('>
<id_list>?
- [ <')'> | <syntax_error: ')' expected> ]
+ <?right_paren>
}
=head2 Macro body
@@ -613,19 +618,22 @@
}
rule pasm_constant {
- <'.constant'> <id> [ <creg> | <constant> ]
+ <'.constant'> <IDENT>
+ [ <pasm_register>
+ | <constant>
+ ]
}
token constant {
- | <float_constant>
- | <int_constant>
- | <string_constant>
+ | <FLOAT_CONST>
+ | <INT_CONST>
+ | <STRING_CONST>
}
# only a PASM register is allowed for pasm_constant
#
-token creg {
+token pasm_register {
[P|I|N|S]\d\d?
}
@@ -643,43 +651,47 @@
}
rule include {
- <'.include'> [ <string_constant> | <syntax_error: string constant
expected> ]
+ <'.include'> <string_constant>
}
rule new_operators {
- <'.pragma'> <'n_operators'> [ <int_constant> | <syntax_error: integer
constant expected> ]
+ <'.pragma'> <'n_operators'> <int_constant>
}
rule loadlib {
- <'.loadlib'> [ <string_constant> | <syntax_error: library name expected> ]
+ <'.loadlib'> <string_constant>
}
rule namespace {
<'.namespace'>
[ <'['>
[ <namespace_id> | <syntax_error: namespace identifier expected> ]
- [ <']'> | <syntax_error: ']' expected> ]
+ <right_bracket>
]?
}
+rule right_bracket {
+ <']'> | <syntax_error: ']' expected>
+}
+
rule hll_specifier {
<'.HLL'>
- [ <string_constant> | <syntax_error: HLL name expected> ]
- [ <','> | <syntax_error: ',' expected> ]
- [ <string_constant> | <syntax_error: HLL type library expected> ]
+ <string_constant>
+ <?comma>
+ <string_constant>
}
rule hll_mapping {
<'.HLL_map'>
- [ <string_constant> | <syntax_error: Parrot Core type expected> ]
- [ <','> | <syntax_error: ',' expected> ]
- [ <string_constant> | <syntax_error: HLL type expected> ]
+ <string_constant>
+ <?comma>
+ <string_constant>
}
rule namespace_id {
- <string_constant>
+ <STRING_CONST>
[ <';'>
- [ <string_constant> | <syntax_error: string constant expected> ]
+ <string_constant>
]*
}
@@ -688,10 +700,10 @@
}
rule id_list {
- <id>
+ <IDENT>
[
<','>
- [ <id> | <syntax_error: identifier expected> ]
+ <id>
]*
}
@@ -701,35 +713,50 @@
=head3 normal_label
-this is a token, because no spaces are allowed between the id and the colon.
+this is a token, because no spaces are allowed between the IDENT and the colon.
=cut
token normal_label {
- <id> <':'>
+ <IDENT> <':'>
}
+rule macro_label {
+ <MACRO_LABEL> | <syntax_error: $LABEL: expected>
+}
-token macro_label {
- <'$'> <id> <':'>
+token MACRO_LABEL {
+ <'$'> <IDENT> <':'>
}
rule macro_label_decl {
- | <'.label'> [ <macro_label> | <syntax_error: $LABEL: expected> ]
+ | <'.label'> <macro_label>
| <normal_label>
}
-token int_constant {
+rule int_constant {
+ <INT_CONST> | <syntax_error: int constant expected>
+}
+
+token INT_CONST {
| 0b\d+
| 0x\d+
| [-]? \d+
}
-token float_constant {
+rule float_constant {
+ <FLOAT_CONST> | <syntax_error: float constant expected>
+}
+
+token FLOAT_CONST {
[-]? \d+\.\d+
}
rule string_constant {
+ <STRING_CONST> | <syntax_error: string constant expected>
+}
+
+rule STRING_CONST {
[ <encoding_specifier>? <charset_specifier> ]?
[ <stringdouble> | <stringsingle> ]
}
@@ -761,8 +788,8 @@
}
token normal_target {
- | <id>
- | <reg>
+ | <IDENT>
+ | <REGISTER>
}
=head2 Macro_target
@@ -772,12 +799,16 @@
=cut
token macro_target {
- | <id>
- | <reg>
+ | <IDENT>
+ | <REGISTER>
| <macro_id>
}
token id {
+ <IDENT> | <syntax_error: identifier expected>
+}
+
+token IDENT {
<!keyword> \w+
}
@@ -785,30 +816,34 @@
<!keyword> \.\w+
}
-token reg {
- | <int_reg>
- | <num_reg>
- | <pmc_reg>
- | <string_reg>
+
+rule register {
+ <REGISTER> | <syntax_error: register expected>
}
+token REGISTER {
+ | <INT_REG>
+ | <NUM_REG>
+ | <PMC_REG>
+ | <STRING_REG>
+}
-token int_reg {
+token INT_REG {
| I\d<[\d]>? # match I0 to I99
| \$I\d+ # match $I0 to infinity
}
-token string_reg {
+token STRING_REG {
| S\d<[\d]>?
| \$S\d+
}
-token pmc_reg {
+token PMC_REG {
| P\d<[\d]>?
| \$P\d+
}
-token num_reg {
+token NUM_REG {
| N\d<[\d]>?
| \$N\d+
}
@@ -829,12 +864,12 @@
=cut
-token nl {
+token NL {
[ \n <?ws> ]+
}
-token NL {
- <nl> | <syntax_error: newline expected>
+token nl {
+ <NL> | <syntax_error: newline expected>
}
regex pod_comment {
