From f741ad7b90cd01c71224b14ced6cc72056c43621 Mon Sep 17 00:00:00 2001
From: Adam Wardynski <award@.(win7-laptop)>
Date: Fri, 29 Apr 2011 09:29:36 +0200
Subject: [PATCH] Parsing WcclFile.
---
libwccl/parser/grammar.g | 222 ++++++++++++++++++++++++++++++++++++++-
1 file changed, 218 insertions(+), 4 deletions(-)
diff --git a/libwccl/parser/grammar.g b/libwccl/parser/grammar.g
index 932d3e1..39b47dc 100644
--- a/libwccl/parser/grammar.g
+++ b/libwccl/parser/grammar.g
@@ -95,6 +95,9 @@ header {
#include <libwccl/ops/match/actions/unmarkmatch.h>
#include <libwccl/ops/functions/match/submatch.h>
+ // Wccl whole file syntax
+ #include <libwccl/wcclfile.h>
+
// Unicode String
#include <unicode/uniset.h>
#include <unicode/unistr.h>
@@ -183,7 +186,7 @@ parse_strset_operator
: body = strset_operator [scope] {
res.reset(new Operator<StrSet>(body, scope.variables()));
}
- EOF
+ (EOF | SEMI)
;
// ----------------------------------------------------------------------------
@@ -199,7 +202,7 @@ parse_bool_operator
: body = bool_operator [scope] {
res.reset(new Operator<Bool>(body, scope.variables()));
}
- EOF
+ (EOF | SEMI)
;
// ----------------------------------------------------------------------------
@@ -215,7 +218,7 @@ parse_symset_operator
: body = symset_operator [scope] {
res.reset(new Operator<TSet>(body, scope.variables()));
}
- EOF
+ (EOF | SEMI)
;
// ----------------------------------------------------------------------------
@@ -231,7 +234,23 @@ parse_position_operator
: body = position_operator [scope] {
res.reset(new Operator<Position>(body, scope.variables()));
}
- EOF
+ (EOF | SEMI)
+;
+
+// ----------------------------------------------------------------------------
+// Rule for parsing match operator with scope.
+// Returns boost::shared_ptr<Operator<Position> >
+parse_match_operator
+ [const Corpus2::Tagset &tagset]
+ returns [boost::shared_ptr<Operator<Match> > res]
+{
+ ParsingScope scope(tagset);
+ boost::shared_ptr<Function<Match> > body;
+}
+ : body = match_operator [scope] {
+ res.reset(new Operator<Match>(body, scope.variables()));
+ }
+ (EOF | SEMI)
;
// ----------------------------------------------------------------------------
@@ -272,6 +291,25 @@ parse_match_rule
: ret_match = match_rule_operator[scope]
;
+// ----------------------------------------------------------------------------
+// ----------------------------------------------------------------------------
+// Rule for parsing wccl files
+parse_wccl_file
+ [const Corpus2::Tagset& tagset]
+ returns [boost::shared_ptr<WcclFile> wccl_file]
+{
+ wccl_file = boost::make_shared<WcclFile>(tagset);
+ boost::shared_ptr<TagRuleSequence> rule_seq;
+}
+ : (any_operator_section [*wccl_file] )*
+ (
+ rule_seq = parse_rule_sequence [tagset] { wccl_file->set_tag_rules(rule_seq); }
+ (any_operator_section [*wccl_file] )*
+ )?
+ EOF
+;
+
+
///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
// VALUES
@@ -1764,6 +1802,139 @@ match_operator
)*
;
+///////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
+// WCCL FILE PARSING RULES
+///////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
+
+any_operator_section
+ [WcclFile& wccl_file]
+{
+ boost::shared_ptr<UntypedOpSequence> untyped_seq;
+ boost::shared_ptr<OpSequence<Bool> > bool_seq;
+ boost::shared_ptr<OpSequence<TSet> > symset_seq;
+ boost::shared_ptr<OpSequence<StrSet> > strset_seq;
+ boost::shared_ptr<OpSequence<Position> > pos_seq;
+ boost::shared_ptr<OpSequence<Match> > m_seq;
+}
+ : untyped_seq = untyped_operator_sequence [wccl_file.tagset()] {
+ wccl_file.add_untyped_section(untyped_seq);
+ }
+ | bool_seq = bool_operator_sequence [wccl_file.tagset()] {
+ wccl_file.add_section(bool_seq);
+ }
+ | symset_seq = symset_operator_sequence [wccl_file.tagset()] {
+ wccl_file.add_section(symset_seq);
+ }
+ | strset_seq = strset_operator_sequence [wccl_file.tagset()] {
+ wccl_file.add_section(strset_seq);
+ }
+ | pos_seq = position_operator_sequence [wccl_file.tagset()] {
+ wccl_file.add_section(pos_seq);
+ }
+ | m_seq = match_operator_sequence [wccl_file.tagset()] {
+ wccl_file.add_section(m_seq);
+ }
+;
+
+bool_operator_sequence
+ [const Corpus2::Tagset& tagset]
+ returns [boost::shared_ptr<OpSequence<Bool> > seq]
+{
+ boost::shared_ptr<Operator<Bool> > op;
+}
+ : BOOL_SECTION_PREFIX name: STRING {
+ seq.reset(new OpSequence<Bool>(token_ref_to_std_string(name)));
+ }
+ LPAREN
+ (op = parse_bool_operator [tagset] { seq->append(op); })+
+ RPAREN
+;
+
+symset_operator_sequence
+ [const Corpus2::Tagset& tagset]
+ returns [boost::shared_ptr<OpSequence<TSet> > seq]
+{
+ boost::shared_ptr<Operator<TSet> > op;
+}
+ : TST_SECTION_PREFIX name: STRING {
+ seq.reset(new OpSequence<TSet>(token_ref_to_std_string(name)));
+ }
+ LPAREN
+ (op = parse_symset_operator [tagset] { seq->append(op); })+
+ RPAREN
+;
+
+strset_operator_sequence
+ [const Corpus2::Tagset& tagset]
+ returns [boost::shared_ptr<OpSequence<StrSet> > seq]
+{
+ boost::shared_ptr<Operator<StrSet> > op;
+}
+ : STR_SECTION_PREFIX name: STRING {
+ seq.reset(new OpSequence<StrSet>(token_ref_to_std_string(name)));
+ }
+ LPAREN
+ (op = parse_strset_operator [tagset] { seq->append(op); })+
+ RPAREN
+;
+
+position_operator_sequence
+ [const Corpus2::Tagset& tagset]
+ returns [boost::shared_ptr<OpSequence<Position> > seq]
+{
+ boost::shared_ptr<Operator<Position> > op;
+}
+ : POS_SECTION_PREFIX name: STRING {
+ seq.reset(new OpSequence<Position>(token_ref_to_std_string(name)));
+ }
+ LPAREN
+ (op = parse_position_operator [tagset] { seq->append(op); })+
+ RPAREN
+;
+
+untyped_operator_sequence
+ [const Corpus2::Tagset& tagset]
+ returns [boost::shared_ptr<UntypedOpSequence> seq]
+{
+ boost::shared_ptr<FunctionalOperator> op;
+}
+ : AT_MARK name: STRING {
+ seq.reset(new UntypedOpSequence(token_ref_to_std_string(name)));
+ }
+ LPAREN
+ (op = any_operator [tagset] { seq->append(op); })+
+ RPAREN
+;
+
+match_operator_sequence
+ [const Corpus2::Tagset& tagset]
+ returns [boost::shared_ptr<OpSequence<Match> > seq]
+{
+ boost::shared_ptr<Operator<Match> > op;
+}
+ : MATCH_SECTION_PREFIX name: STRING {
+ seq.reset(new OpSequence<Match>(token_ref_to_std_string(name)));
+ }
+ LPAREN
+ (op = parse_match_operator [tagset] { seq->append(op); })+
+ RPAREN
+;
+
+any_operator
+ [const Corpus2::Tagset& tagset]
+ returns [boost::shared_ptr<FunctionalOperator> op]
+{
+ static ParsingScope _s(tagset); // just a bogus scope for predicates
+}
+ : (position_operator [_s]) => op = parse_position_operator [tagset]
+ | (symset_operator [_s]) => op = parse_symset_operator [tagset]
+ | (strset_operator [_s]) => op = parse_strset_operator [tagset]
+ | (match_operator [_s]) => op = parse_match_operator [tagset]
+ | op = parse_bool_operator [tagset]
+;
+
// ----------------------------------------------------------------------------
// ----------------------------------------------------------------------------
// Tagging actions and rules:
@@ -2417,6 +2588,41 @@ options {
: "$m:"
;
+STR_SECTION_PREFIX
+options {
+ paraphrase = "Wccl file string operators section prefix";
+}
+ : "@s:"
+;
+
+POS_SECTION_PREFIX
+options {
+ paraphrase = "Wccl file position operators section prefix";
+}
+ : "@p:"
+;
+
+BOOL_SECTION_PREFIX
+options {
+ paraphrase = "Wccl file bool operators section prefix";
+}
+ : "@b:"
+;
+
+TST_SECTION_PREFIX
+options {
+ paraphrase = "Wccl file symset operators section prefix";
+}
+ : "@t:"
+;
+
+MATCH_SECTION_PREFIX
+options {
+ paraphrase = "Wccl file symset operators section prefix";
+}
+ : "@m:"
+;
+
LBRACKET
options {
paraphrase = "'['";
@@ -2466,6 +2672,7 @@ options {
: '@'
;
+
COMMA
options {
paraphrase = "','";
@@ -2487,6 +2694,13 @@ options {
: ':'
;
+SEMI
+options {
+ paraphrase = ";";
+}
+ : ';'
+;
+
SYMBOL
options {
paraphrase = "Symbol";
--
GitLab