From 7992925ba1e5a0092ef0d5f507ed54509ab25414 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Pawe=C5=82=20K=C4=99dzia?= <pawel.kedzia@pwr.wroc.pl>
Date: Fri, 19 Nov 2010 17:00:25 +0100
Subject: [PATCH] Fixed grammar. New operator (equals)
---
libwccl/parser/ANTLRParserResult.h | 21 +
libwccl/parser/Parser.cpp | 41 +-
libwccl/parser/Parser.h | 32 +-
libwccl/parser/grammar.g | 917 ++++++++++++++++-------------
wcclparser/bool_main.cpp | 77 +++
wcclparser/strop_main.cpp | 60 +-
wcclparser/tagset_main.cpp | 78 +++
7 files changed, 746 insertions(+), 480 deletions(-)
create mode 100644 libwccl/parser/ANTLRParserResult.h
create mode 100644 wcclparser/bool_main.cpp
create mode 100644 wcclparser/tagset_main.cpp
diff --git a/libwccl/parser/ANTLRParserResult.h b/libwccl/parser/ANTLRParserResult.h
new file mode 100644
index 0000000..c60f43b
--- /dev/null
+++ b/libwccl/parser/ANTLRParserResult.h
@@ -0,0 +1,21 @@
+#ifndef LIBWCCL_ANTLRPARSERRESULT_H
+#define LIBWCCL_ANTLRPARSERRESULT_H
+
+#include <boost/shared_ptr.hpp>
+
+#include <libwccl/variables.h>
+#include <libwccl/ops/functions.h>
+
+template<class T>
+class ANTLRParserResult
+{
+public:
+ ANTLRParserResult() {
+ this->variables.reset(new Wccl::Variables());
+ }
+
+ boost::shared_ptr<Wccl::Variables> variables;
+ boost::shared_ptr<Wccl::Function<T> > op;
+};
+
+#endif // ANTLRPARSERRESULT_H
diff --git a/libwccl/parser/Parser.cpp b/libwccl/parser/Parser.cpp
index 8f6036e..5bcdb23 100644
--- a/libwccl/parser/Parser.cpp
+++ b/libwccl/parser/Parser.cpp
@@ -22,7 +22,7 @@ Parser::~Parser()
* @arg str writed operator
* @retrun boost::shared_ptr<Wccl::StrSet>
*/
-boost::shared_ptr<Wccl::Function<Wccl::StrSet> > Parser::parseStringOperator(
+boost::shared_ptr<ANTLRParserResult<Wccl::StrSet> > Parser::parseStringOperator(
const std::string& str) const
{
std::stringstream ss (std::stringstream::in | std::stringstream::out);
@@ -32,12 +32,12 @@ boost::shared_ptr<Wccl::Function<Wccl::StrSet> > Parser::parseStringOperator(
}
/**
- * @desc Parse string operator. Runs parse_string_operator rule
+ * @desc Parse string operator. Runs parse_string_operator rule
* in the parser grammar.
* @arg istr input stream with writed operator
* @return boost::shared_ptr<Wccl::Function<Wccl::StrSet> > to created operator
*/
-boost::shared_ptr<Wccl::Function<Wccl::StrSet> > Parser::parseStringOperator(
+boost::shared_ptr<ANTLRParserResult<Wccl::StrSet> > Parser::parseStringOperator(
std::istream& istr) const
{
ANTLRLexer lexer(istr);
@@ -53,7 +53,7 @@ boost::shared_ptr<Wccl::Function<Wccl::StrSet> > Parser::parseStringOperator(
* @arg str writed predicate(s)
* @return boost::shared_ptr<Wccl::Function<Wccl::Bool> > to created predicate
*/
-boost::shared_ptr<Wccl::Function<Wccl::Bool> > Parser::parsePredicate(
+boost::shared_ptr<ANTLRParserResult<Wccl::Bool> > Parser::parsePredicate(
const std::string& str) const
{
std::stringstream ss (std::stringstream::in | std::stringstream::out);
@@ -67,7 +67,7 @@ boost::shared_ptr<Wccl::Function<Wccl::Bool> > Parser::parsePredicate(
* @arg istr input stream with writed predicate
* @return boost::shared_ptr<Wccl::Function<Wccl::Bool> > to created predicate
*/
-boost::shared_ptr<Wccl::Function<Wccl::Bool> > Parser::parsePredicate(
+boost::shared_ptr<ANTLRParserResult<Wccl::Bool> > Parser::parsePredicate(
std::istream& istr) const
{
ANTLRLexer lexer(istr);
@@ -77,33 +77,32 @@ boost::shared_ptr<Wccl::Function<Wccl::Bool> > Parser::parsePredicate(
}
// ----------------------------------------------------------------------------
-
/**
- * @desc Parse values writed in std::string. Converts writed values
- * to istream and calling parseValue with istream
- * @arg str writed value(s)
- * @retrun boost::shared_ptr<Wccl::Value>
+ * @desc Parse sym set operator writed in std::string. Converts writed operator
+ * to istream and calling parseSymSetOperator with istream
+ * @arg str writed operator
+ * @retrun boost::shared_ptr<Wccl::Function<Wccl::TSet> >
*/
-/*
-boost::shared_ptr<Wccl::Value> Parser::parseValue(const std::string& str) const
+boost::shared_ptr<ANTLRParserResult<Wccl::TSet> > Parser::parseSymSetOperator(
+ const std::string& str) const
{
std::stringstream ss (std::stringstream::in | std::stringstream::out);
ss << str;
- return this->parseValue(ss);
+ return this->parseSymSetOperator(ss);
}
-*/
+
/**
- * @desc Parse values. Runs parse_values rule in the parser grammar.
- * @arg istr input stream with writed values
- * @return boost::shared_ptr<Wccl::Value> to created value
+ * @desc Parse sym set operator. Runs parse_sym_set_operator rule
+ * in the parser grammar.
+ * @arg istr input stream with writed operator
+ * @return boost::shared_ptr<Wccl::Function<Wccl::TSet> > to created operator
*/
-/*
-boost::shared_ptr<Wccl::Value> Parser::parseValue(std::istream& istr) const
+boost::shared_ptr<ANTLRParserResult<Wccl::TSet> > Parser::parseSymSetOperator(
+ std::istream& istr) const
{
ANTLRLexer lexer(istr);
ANTLRParser parser(lexer);
- return parser.parse_values();
+ return parser.parse_sym_set_operator();
}
-*/
diff --git a/libwccl/parser/Parser.h b/libwccl/parser/Parser.h
index 4b0177b..1cbb11c 100644
--- a/libwccl/parser/Parser.h
+++ b/libwccl/parser/Parser.h
@@ -4,6 +4,8 @@
#include "ANTLRLexer.hpp"
#include "ANTLRParser.hpp"
+#include <iostream>
+
// ----------------------------------------------------------------------------
#include <sstream>
@@ -16,6 +18,7 @@
// exceptions
#include <libwccl/parser/ParserException.h>
+#include <libwccl/parser/ANTLRParserResult.h>
// ----------------------------------------------------------------------------
@@ -25,16 +28,25 @@ public:
~Parser();
// ---------------------------------------------------------------------------
- // methods for parsing string operator
- boost::shared_ptr<Wccl::Function<Wccl::StrSet> >
- parseStringOperator(const std::string&) const;
- boost::shared_ptr<Wccl::Function<Wccl::StrSet> >
- parseStringOperator(std::istream& ) const;
- // methods for parsing predicates (returns bool)
- boost::shared_ptr<Wccl::Function<Wccl::Bool> >
- parsePredicate(const std::string&) const;
- boost::shared_ptr<Wccl::Function<Wccl::Bool> >
- parsePredicate(std::istream& ) const;
+ // methods for parsing string operators
+ boost::shared_ptr<ANTLRParserResult<Wccl::StrSet> >
+ parseStringOperator(const std::string&) const;
+ boost::shared_ptr<ANTLRParserResult<Wccl::StrSet> >
+ parseStringOperator(std::istream&) const;
+
+ // ---------------------------------------------------------------------------
+ // methods for parsing bool operators
+ boost::shared_ptr<ANTLRParserResult<Wccl::Bool> >
+ parsePredicate(const std::string&) const;
+ boost::shared_ptr<ANTLRParserResult<Wccl::Bool> >
+ parsePredicate(std::istream&) const;
+
+ // ---------------------------------------------------------------------------
+ // methods for parsing bool operators
+ boost::shared_ptr<ANTLRParserResult<Wccl::TSet> >
+ parseSymSetOperator(const std::string&) const;
+ boost::shared_ptr<ANTLRParserResult<Wccl::TSet> >
+ parseSymSetOperator(std::istream&) const;
private:
const Corpus2::Tagset &tagset;
diff --git a/libwccl/parser/grammar.g b/libwccl/parser/grammar.g
index 7818058..82a3877 100644
--- a/libwccl/parser/grammar.g
+++ b/libwccl/parser/grammar.g
@@ -21,15 +21,24 @@ header {
#include <libwccl/ops/nor.h>
#include <libwccl/ops/and.h>
#include <libwccl/ops/affix.h>
+ #include <libwccl/ops/regex.h>
+ #include <libwccl/ops/equals.h>
#include <libwccl/ops/toupper.h>
#include <libwccl/ops/tolower.h>
#include <libwccl/ops/constant.h>
#include <libwccl/ops/functions.h>
+ #include <libwccl/ops/vargetter.h>
+ #include <libwccl/ops/varsetter.h>
+ // #include <libwccl/ops/intersects.h>
+ // #include <libwccl/ops/issubsetof.h>
#include <libwccl/ops/logicalpredicate.h>
// Unicode String
#include <unicode/uniset.h>
#include <unicode/unistr.h>
+
+ //
+ #include "ANTLRParserResult.h"
}
options {
@@ -60,78 +69,75 @@ private:
int token_ref_to_int(antlr::RefToken& rstr) {
return atoi(((antlr::Token*)rstr)->getText().c_str());
}
+}
- // hepls function for processing
- boost::shared_ptr<Wccl::Function<Wccl::StrSet> > get_str_set_expr(
- boost::shared_ptr<Wccl::StrSet> ret_str_set)
- {
- boost::shared_ptr<Wccl::Function<Wccl::StrSet> > strset_expr(
- new Wccl::Constant<Wccl::StrSet>(*ret_str_set.get())
- );
+// TODO
+// - nie mozna utworzyc in/inter/equal z operatorem?
+// - jak utworzyc TSet (dodac element do TSet)
+// - nie mozna utworzy Const::Value i na tym robic specjalizowany reset?
+// - base, orth
- return strset_expr;
- }
+// Bylo boolean_v TRUE_VALUE. FALSE_VALE, a jest "True", "False" - bylo z
+// regula lexera, a teraz jest wstawione na "sztywno" do gramatyki
- Wccl::SentenceContext get_tmp_context() {
- boost::shared_ptr<Corpus2::Sentence> sentence;
- Wccl::SentenceContext sc(sentence);
-
- return sc;
- }
-}
///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
-// ----------------------------------------------------------------------------------
-// Start all rules
-/*
-start_rules
-{
- std::string name = "";
-}
- : values_ref [name] { fprintf(stderr, "%s\n", name.c_str()); }
- | position_op [name] { fprintf(stderr, "%s\n", name.c_str()); }
- | filters_op [name] { fprintf(stderr, "%s\n", name.c_str()); }
- | setvar_op [name] { fprintf(stderr, "%s\n", name.c_str()); }
- | boolean_op [name] { fprintf(stderr, "%s\n", name.c_str()); }
- ;
-*/
-
+// "GLOBAL" RULES
///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
-// "GLOBAL" RULES
// ----------------------------------------------------------------------------
// Rules for parsing string operators in scope (variables).
// Returns boost::shared_ptr<Wccl::Function<Wccl::StrSet> >
-parse_string_operator
- returns [boost::shared_ptr<Wccl::Function<Wccl::StrSet> > res]
+parse_string_operator
+ returns [boost::shared_ptr<ANTLRParserResult<Wccl::StrSet> > res]
{
- Wccl::Variables vars;
+ res.reset(new ANTLRParserResult<Wccl::StrSet>());
+ boost::shared_ptr<Wccl::Function<Wccl::StrSet> > op;
}
- : res = string_operators [vars]
+ : op = string_operators [*res->variables.get()] {
+ res->op = op;
+ }
;
+
// ----------------------------------------------------------------------------
// Rules for parsing predicates in scope (variables).
// Returns boost::shared_ptr<Wccl::Function<Wccl::Bool> >
parse_predicates
- returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > res]
+ returns [boost::shared_ptr<ANTLRParserResult<Wccl::Bool> > res]
{
- Wccl::Variables vars;
+ res.reset(new ANTLRParserResult<Wccl::Bool>());
+ boost::shared_ptr<Wccl::Function<Wccl::Bool> > op;
}
- : res = predicates [vars]
+ : op = predicates [*res->variables.get()] {
+ res->op = op;
+ }
;
+
// ----------------------------------------------------------------------------
// Rules for parsing values in scope (variables).
-// Returns boost::shared_ptr<Wccl::Value>
+// Returns boost::shared_ptr<Wccl::Function<Wccl::Value> >
/*
parse_values
- returns [boost::shared_ptr<Wccl::Value> ret]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Value> > res]
{
Wccl::Variables vars;
}
- : ret = values [vars]
+ : res = values [vars]
;
*/
-
+// ----------------------------------------------------------------------------
+// Rules for parsing tagset (symbol set) operators
+// Returns boost::shared_ptr<Wccl::Function<Wccl::TSet> >
+parse_sym_set_operator
+ returns [boost::shared_ptr<ANTLRParserResult<Wccl::TSet> > res]
+{
+ res.reset(new ANTLRParserResult<Wccl::TSet>());
+ boost::shared_ptr<Wccl::Function<Wccl::TSet> > op;
+}
+ : op = sym_set_operators [*res->variables.get()] {
+ res->op = op;
+ }
+;
///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
// All values:
@@ -140,146 +146,220 @@ parse_values
/*
values
[Wccl::Variables& vars]
- returns [boost::shared_ptr<Wccl::Constant<Wccl::Value> > res]
- : res = position [vars]
- | res = str_set [vars]
- | res = sym_set [vars]
- | res = boolean [vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Value> > res]
+{
+ boost::shared_ptr<Wccl::Constant<Wccl::StrSet> > r_StrSet;
+ boost::shared_ptr<Wccl::Constant<Wccl::TSet> > r_TSet;
+ boost::shared_ptr<Wccl::Constant<Wccl::Bool> > r_Bool;
+ boost::shared_ptr<Wccl::Constant<Wccl::Position> > r_Pos;
+}
+ : r_Pos = position [vars] // TODO { res = r_Pos; }
+ | r_TSet = sym_set [vars]
+ | r_Bool = boolean [vars]
;
*/
+///////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
+// VALUES
+///////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
// ----------------------------------------------------------------------------
-// Values reference => values + position_ref
-// !! Cannot use for setvar(...,...) !!
-/*
-values_ref [std::string& name]:
- values [name]
- | position_ref [name]
- | boolean_ref [name]
- ;
-*/
+// Single or muliple values in string set
+str_set_v_in
+ [boost::shared_ptr<Wccl::StrSet>& s_set]
+ : v1: STRING {
+ s_set->insert(token_ref_to_ustring(v1));
+ }
+ | v2: STRING COMMA str_set_v_in [s_set] {
+ s_set->insert(token_ref_to_ustring(v2));
+ }
+;
+// string set, called as unnamed (temporary) StrSet:
+// calls: [] ['a'] ['a', 'b'] ["a"] ["a", "b"] ['a', "b"] or variable $A
+str_set_v
+ returns [boost::shared_ptr<Wccl::Constant<Wccl::StrSet> > val]
+{
+ boost::shared_ptr<Wccl::StrSet> set(new Wccl::StrSet);
+}
+ : LBRACKET RBRACKET {
+ val.reset(new Wccl::Constant<Wccl::StrSet>(*set.get()));
+ }
+ | LBRACKET str_set_v_in [set] RBRACKET {
+ val.reset(new Wccl::Constant<Wccl::StrSet>(*set.get()));
+ }
+;
+// ----------------------------------------------------------------------------
+// element of sym set
+sym_set_elem
+ [boost::shared_ptr<Wccl::TSet> &t_set]
+ : s1: SYMBOL {
+// t_set->insert(token_ref_to_ustring(s1));
+ }
+ | G_MARK s2: SYMBOL G_MARK {
+// t_set->insert(token_ref_to_ustring(s2));
+ }
+
+ | s3: SYMBOL COMMA sym_set_elem [t_set] {
+// t_set->insert(token_ref_to_ustring(s3));
+ }
+ | G_MARK s4: SYMBOL G_MARK COMMA sym_set_elem [t_set] {
+// t_set->insert(token_ref_to_ustring(s3));
+ }
+;
+
+// sym set in
+sym_set_in
+ [boost::shared_ptr<Wccl::TSet> &set]
+ : sym_set_elem [set]
+;
+// sym set {} {a} {a, b}
+sym_set_v
+ returns [boost::shared_ptr<Wccl::Constant<Wccl::TSet> > val]
+{
+ boost::shared_ptr<Wccl::TSet> set(new Wccl::TSet);
+}
+ : LCURLY RCURLY {
+ val.reset(new Wccl::Constant<Wccl::TSet>(*set.get()));
+ }
+ | LCURLY sym_set_in [set] RCURLY {
+ val.reset(new Wccl::Constant<Wccl::TSet>(*set.get()));
+ }
+;
+// ----------------------------------------------------------------------------
+// boolean value:
+boolean_v
+ returns [boost::shared_ptr<Wccl::Constant<Wccl::Bool> > val]
+ : TRUE_VALUE { val.reset(new Wccl::Constant<Wccl::Bool>(Wccl::Bool(true ))); }
+ | FALSE_VALUE { val.reset(new Wccl::Constant<Wccl::Bool>(Wccl::Bool(false))); }
+;
+// ----------------------------------------------------------------------------
+// position value:
+position_v
+ returns [boost::shared_ptr<Wccl::Constant<Wccl::Position> > val]
+ : i: INT {
+ val.reset(
+ new Wccl::Constant<Wccl::Position>(Wccl::Position(token_ref_to_int(i)))
+ );
+ }
+ | "begin" {
+ val.reset(
+ new Wccl::Constant<Wccl::Position>(
+ Wccl::Position(Wccl::Position::Begin)
+ )
+ );
+ }
+ | "end" {
+ val.reset(
+ new Wccl::Constant<Wccl::Position>(
+ Wccl::Position(Wccl::Position::End)
+ )
+ );
+ }
+ | "nowhere" {
+ val.reset(
+ new Wccl::Constant<Wccl::Position>(
+ Wccl::Position(Wccl::Position::Nowhere)
+ )
+ );
+ }
+;
+
+///////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
+// VARIABLES
///////////////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////////////
-// Position: $0name
+// Position: $name
// ----------------------------------------------------------------------------
-// TODO Cos nie lapie dobrze implementacja!!! Moze jakas dodatkowa
-// TODO regula do lexera?
position
[Wccl::Variables& vars]
- returns [boost::shared_ptr<Wccl::Constant<Wccl::Position> > op]
-{
- boost::shared_ptr<Wccl::Position> val;
-}
- : DOLLAR "0" n: SYMBOL {
- val = vars.get_put<Wccl::Position>(token_ref_to_std_string(n));
- op.reset(new Wccl::Constant<Wccl::Position>(*val.get()));
+ returns [boost::shared_ptr<Wccl::VarGetter<Wccl::Position> > op]
+ : DOLLAR n: VAR_NAME {
+ vars.get_put<Wccl::Position>(token_ref_to_std_string(n));
+
+ Wccl::VariableAccesor<Wccl::Position> acc =
+ vars.create_accesor<Wccl::Position>(token_ref_to_std_string(n));
+
+ op.reset(new Wccl::VarGetter<Wccl::Position>(acc));
}
;
// ----------------------------------------------------------------------------
// Position reference: $(0-9)+name
// !! Cannot use for setvar(...,...) !!
-/*
position_ref
[Wccl::Variables& vars]
- returns [boost::shared_ptr<Wccl::PositionRef> val]
- : DOLLAR p_ref: INT n: SYMBOL {
- val.reset(
- new Wccl::PositionRef(
- vars.get_put<Wccl::Position>(token_ref_to_std_string(n)),
- token_ref_to_int(p_ref)
- )
- );
+ : DOLLAR p_ref: INT n: VAR_NAME {
+ // TODO
}
;
-*/
// ----------------------------------------------------------------------------
-// String set, call examples: $name, $Name, $_name, $_Name etc.
+// String set, call examples: $name, $sName, $s_name, $s_Name etc.
// This expression gets variable of tyme StrSet from string-named variable
-// Returns Wccl::Function<Wccl::StrSet> from Set-variables
+// Returns Wccl::VarGetter<Wccl::StrSet> from Set-variables
str_set
[Wccl::Variables& vars]
- returns [boost::shared_ptr<Wccl::Constant<Wccl::StrSet> > op]
-{
- boost::shared_ptr<Wccl::StrSet> val;
-}
- : DOLLAR n: SYMBOL {
- val = vars.get_put<Wccl::StrSet>(token_ref_to_std_string(n));
- op.reset(new Wccl::Constant<Wccl::StrSet>(*val.get()));
+ returns [boost::shared_ptr<Wccl::VarGetter<Wccl::StrSet> > op]
+ : DOLLAR STR_PREFIX n: VAR_NAME {
+ // get/put variable to variables
+ vars.get_put<Wccl::StrSet>(token_ref_to_std_string(n));
+
+ // makes accesor for value
+ Wccl::VariableAccesor<Wccl::StrSet> acc =
+ vars.create_accesor<Wccl::StrSet>(token_ref_to_std_string(n));
+
+ op.reset(new Wccl::VarGetter<Wccl::StrSet>(acc));
}
;
// ----------------------------------------------------------------------------
-// Symbol set: $$name
+// Symbol set: $tName
sym_set
[Wccl::Variables& vars]
- returns [boost::shared_ptr<Wccl::Constant<Wccl::TSet> > op]
-{
- boost::shared_ptr<Wccl::TSet> val;
-}
- : DOLLAR DOLLAR n: SYMBOL {
- val = vars.get_put<Wccl::TSet>(token_ref_to_std_string(n));
- op.reset(new Wccl::Constant<Wccl::TSet>(*val.get()));
+ returns [boost::shared_ptr<Wccl::VarGetter<Wccl::TSet> > op]
+ : DOLLAR TST_PREFIX n: VAR_NAME {
+ vars.get_put<Wccl::TSet>(token_ref_to_std_string(n));
+
+ Wccl::VariableAccesor<Wccl::TSet> acc =
+ vars.create_accesor<Wccl::TSet>(token_ref_to_std_string(n));
+
+ op.reset(new Wccl::VarGetter<Wccl::TSet>(acc));
}
;
// ----------------------------------------------------------------------------
-// Bool: $?name
+// Bool: $bName
boolean
[Wccl::Variables& vars]
- returns [boost::shared_ptr<Wccl::Constant<Wccl::Bool> > op]
-{
- boost::shared_ptr<Wccl::Bool> val;
-}
- : DOLLAR Q_MARK n: SYMBOL {
- val = vars.get_put<Wccl::Bool>(token_ref_to_std_string(n));
- op.reset(new Wccl::Constant<Wccl::Bool>(*val.get()));
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > op]
+ : DOLLAR BOOL_PREFIX n: VAR_NAME {
+ vars.get_put<Wccl::Bool>(token_ref_to_std_string(n));
+
+ Wccl::VariableAccesor<Wccl::Bool> acc =
+ vars.create_accesor<Wccl::Bool>(token_ref_to_std_string(n));
+
+ op.reset(new Wccl::VarGetter<Wccl::Bool>(acc));
}
;
-// Boolean $!name
-/*
-boolean_ref [std::string& name]:
- DOLLAR E_MARK n1: SYMBOL { name = token_ref_to_std_string(n1); }
- ;
-*/
-/////////////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
// OPERATORS
-/////////////////////////////////////////////////////////////////////////////////////
-// ----------------------------------------------------------------------------------
-// Positions operator
-// TODO range przyjmuje postion_ref. ?? Zmienic na position ??
-/*
-position_op [std::string& name]
-{
- std::string r1, r2;
-}
- : "flex" LBRACKET position_ref [name] RBRACKET
- | "range" LPAREN s: SYMBOL COMMA position_ref [r1] COMMA position_ref [r2] RPAREN
- { name = ("Range opertator from " + token_ref_to_std_string(s) + " [" + r1 + ":" + r2 + "]!"); }
- ;
-*/
-// ----------------------------------------------------------------------------------
-// Filtering operator
-/*
-filters_op [std::string& name]
-{
- std::string p, p2, e1, e2;
-}
- : "catflt" LPAREN position_ref [p] COMMA es_any [e1] COMMA es_any [e2] RPAREN
- { name = ( "Catflt operator in position " + p + " for sets " + e1 + " " + e2); }
- | "agrflt" LPAREN position_ref [p] COMMA position_ref [p2] COMMA es_any [e1] COMMA i: INT RPAREN
- { name = ( "Agrflt operator p1 " + p + " p2 " + p2 + " for set " + e1 + " aggr_attrs " + token_ref_to_std_string(i)); }
- ;
-*/
-
-// ----------------------------------------------------------------------------------
+///////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
+// ----------------------------------------------------------------------------
+// ----------------------------------------------------------------------------
// Setvar operator
+// ----------------------------------------------------------------------------
/*
-setvar_op [std::string& value]
+setvar_op
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > ret]
: setvar_pos [value]
| setvar_bool [value]
- | setvar_sset [value]
+ : ret = setvar_sset [vars]
| setvar_tset [value]
- ;
+;
*/
-
+// Implementations of setvar:
+// ----------------------------------------------------------------------------
// setvar dla position przyjmuje position_ref -> TODO sprawdzic dlaczego
// gramatyka nie pokrywa "setvar" LPAREN position COMMA position_v RPAREN
/*
@@ -293,10 +373,22 @@ setvar_bool [std::string& value]
: "setvar" LPAREN boolean [value] COMMA boolean_v [value] RPAREN
;
*/
+
/*
-setvar_sset [std::string& value]
- : "setvar" LPAREN str_set [value] COMMA str_set_v [value] RPAREN
- ;
+setvar_sset
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::VarSetter<Wccl::Bool> > op]
+{
+ boost::shared_ptr<Wccl::VarGetter<Wccl::StrSet> > ret_str_var;
+ boost::shared_ptr<Wccl::Function<Wccl::StrSet> > ret_str_op;
+}
+ : "setvar" LPAREN ret_str_var = str_set [vars] COMMA
+ ret_str_op = string_operators [vars] RPAREN {
+// VariableAccesor<Bool> acc_t = cx.variables()->create_accesor<Bool>("True_bool");
+// VarSetter<Bool> var_setter(acc_t, false_constant);
+// Wccl::VariableAccesor<Wccl::StrSet> acc_t = vars->create_accesor<Wccl::StrSet>("True_bool");
+ }
+;
*/
/*
setvar_tset [std::string& value]
@@ -304,247 +396,86 @@ setvar_tset [std::string& value]
;
*/
-///////////////////////////////////////////////////////////////////////////////
-///////////////////////////////////////////////////////////////////////////////
-// VALUES
-///////////////////////////////////////////////////////////////////////////////
// ----------------------------------------------------------------------------
-// Single or muliple values in string set
-str_set_v_in
- [boost::shared_ptr<Wccl::StrSet>& s_set]
- : v1: STRING { s_set->insert(token_ref_to_ustring(v1)); }
- | v2: STRING COMMA str_set_v_in [s_set] {
- s_set->insert(token_ref_to_ustring(v2));
- }
+// ----------------------------------------------------------------------------
+// Symbol set (tagset) operators
+// Returns boost::shared_ptr<Wccl::Function<Wccl::TSet> >
+// ----------------------------------------------------------------------------
+sym_set_operators
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::TSet> > ret]
+ : ret = op_sym_set [vars]
;
-// string set, called as unnamed (temporary) StrSet:
-// calls: [] ['a'] ['a', 'b'] ["a"] ["a", "b"] ['a', "b"] or variable $A
-str_set_v
+// Implementations of symbol set operators:
+// ----------------------------------------------------------------------------
+op_sym_set
[Wccl::Variables& vars]
- returns [boost::shared_ptr<Wccl::Constant<Wccl::StrSet> > val]
-{
- boost::shared_ptr<Wccl::StrSet> set(new Wccl::StrSet);
-}
- : LBRACKET RBRACKET {
- val.reset(new Wccl::Constant<Wccl::StrSet>(*set.get()));
- }
- | LBRACKET str_set_v_in [set] RBRACKET {
- val.reset(new Wccl::Constant<Wccl::StrSet>(*set.get()));
- }
- | val = str_set [vars]
- ;
+ returns [boost::shared_ptr<Wccl::Function<Wccl::TSet> > op]
+ : op = sym_set [vars]
+ | op = sym_set_v
+;
+
// ----------------------------------------------------------------------------
-// element of sym set
-/*
-sym_set_elem_s [std::string& value]
- : s1: SYMBOL { value += token_ref_to_std_string(s1); }
- | s2: SYMBOL COMMA sym_set_elem_s [value] { value += token_ref_to_std_string(s2); }
- | s3: SYMBOL COMMA sym_set_elem_g [value] { value += token_ref_to_std_string(s3); }
- ;
-*/
-// element of sym set
-/*
-sym_set_elem_g [std::string& value]
- : G_MARK s1: SYMBOL G_MARK { value += token_ref_to_std_string(s1); }
- | G_MARK s2: SYMBOL G_MARK COMMA sym_set_elem_g [value] { value += token_ref_to_std_string(s2); }
- | G_MARK s3: SYMBOL G_MARK COMMA sym_set_elem_s [value] { value += token_ref_to_std_string(s3); }
- ;
-*/
-// sym set in
-/*
-sym_set_in [std::string& value]
- : sym_set_elem_s [value]
- | sym_set_elem_g [value]
- ;
-*/
-// sym set {} {a} {a, b}
-/*
-sym_set_v [std::string& value]
- : LCURLY RCURLY
- | LCURLY sym_set_in [value] RCURLY
- ;
-*/
// ----------------------------------------------------------------------------
-// boolean:
-boolean_v
- [Wccl::Variables& vars]
- returns [boost::shared_ptr<Wccl::Constant<Wccl::Bool> > val]
- : "True" { val.reset(new Wccl::Constant<Wccl::Bool>(Wccl::Bool(true ))); }
- | "False" { val.reset(new Wccl::Constant<Wccl::Bool>(Wccl::Bool(false))); }
- | val = boolean [vars]
-;
+// Position operators
+// Returns boost::shared_ptr<Wccl::Function<Wccl::Position> >
// ----------------------------------------------------------------------------
-// position value:
-/*
-position_v [std::string& value]
- : i: INT { value = token_ref_to_std_string(i); }
- | "begin" { value = "begin"; }
- | "end" { value = "end"; }
- | "nowhere" { value = "nowhere"; }
- | position [value]
- ;
-*/
+position_operators
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Position> > ret]
+ : ret = op_position [vars]
+;
+// Implementations of symbol set operators:
// ----------------------------------------------------------------------------
-// internal values:
-/*
-v_literal [std::string& value]
- : s1: STRING { value = token_ref_to_std_string(s1); }
- | s2: SYMBOL { value = token_ref_to_std_string(s2); }
- ;
-*/
-/////////////////////////////////////////////////////////////////////////////////////
-// constants
-// set of values
-/*
-st::shared_ptr<Wccl::StrSet> ret]s_literal [std::string& v]
- : LBRACKET ((v_literal[v]) (COMMA v_literal [v])*)? RBRACKET
- ;
-*/
-
-// comma-separated predicates
-/*
-seq_et [std::string& v]:
- et_any [v] (COMMA et_any [v])*
- ;
-*/
-/*
-es_any [Wccl::Variables& vars] returns [boost::shared_ptr<Wccl::StrSet> ret]
- s_literal [v]
- | es_op [v]
- ;
-*/
-/*
-et_bool [std::string& v]:
- boolean [v]
- | boolean_op [v]
- ;
-*/
-
-// set relations
-/*
-et_set [std::string& v]
-{
- std::string v1, v2;
-}
- : "in" LPAREN es_any [v1] COMMA es_any [v2] RPAREN { v = ("name " + v1 + " " + v2); }
- | "inter" LPAREN es_any [v1] COMMA es_any [v2] RPAREN { v = ("inter " + v1 + " " + v2); }
- | "equal" LPAREN es_any [v1] COMMA es_any [v2] RPAREN { v = ("equal " + v1 + " " + v2); }
- ;
-*/
-/*
-et_string [std::string& v]
- : "isbig" LPAREN es_any [v] RPAREN
- | "hasnum" LPAREN es_any [v] RPAREN
- ;
-*/
-/*
-et_action [std::string& v]
-{
- std::string v1, v2;
-}
- : "delete" LPAREN et_any [v] RPAREN
- | "select" LPAREN et_any [v] RPAREN
- | "relabel" LPAREN es_any [v1] COMMA et_any [v2] RPAREN { v = ("relabel " + v1 + " " + v2); }
- | "unify" LPAREN es_any [v1] COMMA i: INT RPAREN { v = ("relabel " + v1 + " on position " + token_ref_to_std_string(i)); }
- | "mark" LPAREN s1: SYMBOL RPAREN { v = ("mark " + token_ref_to_std_string(s1)); }
- | "unmark" LPAREN s2: SYMBOL RPAREN { v = ("unmark " + token_ref_to_std_string(s2)); }
- | "startnew" LPAREN s3: SYMBOL RPAREN { v = ("startnew " + token_ref_to_std_string(s3)); }
- | "lextend" LPAREN s4: SYMBOL RPAREN { v = ("lextend " + token_ref_to_std_string(s4)); }
- | "rextend" LPAREN s5: SYMBOL RPAREN { v = ("rextend " + token_ref_to_std_string(s5)); }
- ;
-*/
-/*
-et_iter [std::string& v]
-{
- std::string v1, v2, v3, v4;
-}
- : "only" LPAREN position_ref [v1] COMMA position_ref [v2] COMMA position_ref [v3] COMMA et_any [v4] RPAREN
- | "atleast" LPAREN position_ref [v1] COMMA position_ref [v2] COMMA position_ref [v3] COMMA et_any [v4] COMMA i:INT RPAREN
- | "llook" LPAREN position_ref [v1] COMMA position_ref [v2] COMMA position_ref [v3] COMMA et_any [v4] RPAREN
- | "rlook" LPAREN position_ref [v1] COMMA position_ref [v2] COMMA position_ref [v3] COMMA et_any [v4] RPAREN
- | "setvar" LPAREN position_ref [v1] COMMA position_ref [v2] RPAREN
- | "lskip" LPAREN position_ref [v1] COMMA SYMBOL COMMA position_ref [v2] COMMA et_any [v3] RPAREN
- | "lphrase" LPAREN position_ref [v1] COMMA SYMBOL COMMA position_ref [v2] RPAREN
- | "rphrase" LPAREN position_ref [v1] COMMA SYMBOL COMMA position_ref [v2] RPAREN
- | "accept" LPAREN seq_et [v1] RPAREN
- ;
-*/
-
-// predicates checking agreement
-/*
-et_agr [std::string& name]
-{
- std::string p1, p2, v;
-}
- : "agr" LPAREN position_ref [p1] COMMA position_ref [p2] COMMA es_any [v] COMMA i1: INT RPAREN
- | "agrpp" LPAREN position_ref [p1] COMMA position_ref [p2] COMMA es_any [v] COMMA i2: INT RPAREN
- | "wagr" LPAREN position_ref [p1] COMMA position_ref [p2] COMMA es_any [v] COMMA i3: INT RPAREN
- ;
-*/
-/*
-// annotation checking predicates
-et_annot [std::string& v]
- : "phrase" LPAREN position_ref [v] COMMA s: SYMBOL RPAREN
- ;
-*/
-/*
-// constraints
-et_any [std::string& v]
- : et_bool [v]
- | et_set [v]
- | et_string [v]
- | et_action [v]
- | et_iter [v]
- | et_agr [v]
- | et_annot [v]
- ;
-*/
+op_position
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Position> > op]
+ : op = position [vars]
+ | op = position_v
+;
-///////////////////////////////////////////////////////////////////////////////
-///////////////////////////////////////////////////////////////////////////////
-// Stiring operators returns boost::shared_ptr<Wccl::Function<Wccl::StrSet> >
-///////////////////////////////////////////////////////////////////////////////
+// ----------------------------------------------------------------------------
+// ----------------------------------------------------------------------------
+// Stiring operators
+// Returns boost::shared_ptr<Wccl::Function<Wccl::StrSet> >
// ----------------------------------------------------------------------------
string_operators
[Wccl::Variables& vars]
returns [boost::shared_ptr<Wccl::Function<Wccl::StrSet> > ret]
- /*
- : ret = op_orth [vars]
- | ret = op_base [vars]
- */
- : ret = op_lower [vars]
- | ret = op_upper [vars]
- | ret = op_affix [vars]
- | ret = str_set_v [vars]
+ : ret = op_orth [vars]
+ | ret = op_base [vars]
+ | ret = op_lower [vars]
+ | ret = op_upper [vars]
+ | ret = op_affix [vars]
+ | ret = op_str_set [vars]
;
// Implementations of string operators:
// ----------------------------------------------------------------------------
-/*
-op_orth [Wccl::Variables& vars] returns [boost::shared_ptr<Wccl::StrSet> ret]
+op_orth
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::StrSet> > ret]
{
- boost::shared_ptr<Wccl::PositionRef> tmpPosRef;
+ // TODO
}
- : "orth" LBRACKET tmpPosRef = position_ref [vars] RBRACKET {
- // TODO
+ : "orth" LBRACKET position_ref [vars] RBRACKET {
+ // ret = TODO
}
;
-*/
// ----------------------------------------------------------------------------
-/*
-op_base [Wccl::Variables& vars] returns [boost::shared_ptr<Wccl::StrSet> ret]
+op_base
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::StrSet> > ret]
{
- boost::shared_ptr<Wccl::PositionRef> tmpPosRef;
+ // TODO
}
- : "base" LBRACKET tmpPosRef = position_ref [vars] RBRACKET {
- // TODO
+ : "base" LBRACKET position_ref [vars] RBRACKET {
+ // ret = TODO
}
;
-*/
// ----------------------------------------------------------------------------
-// returns boost::shared_ptr<Wccl::Function<Wccl::StrSet> >
op_lower
- [Wccl::Variables& vars] returns
- [boost::shared_ptr<Wccl::Function<Wccl::StrSet> > ret]
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::StrSet> > ret]
{
boost::shared_ptr<Wccl::Function<Wccl::StrSet> > o_ret;
}
@@ -574,11 +505,19 @@ op_affix
ret.reset(new Wccl::Affix(o_ret, token_ref_to_int(offset)));
}
;
-///////////////////////////////////////////////////////////////////////////////
-///////////////////////////////////////////////////////////////////////////////
-// Predicates returns boost::shared_ptr<Wccl::Function<Wccl::Bool> >
-///////////////////////////////////////////////////////////////////////////////
-///////////////////////////////////////////////////////////////////////////////
+// ----------------------------------------------------------------------------
+op_str_set
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::StrSet> > op]
+ : op = str_set [vars]
+ | op = str_set_v
+;
+
+// ----------------------------------------------------------------------------
+// ----------------------------------------------------------------------------
+// Logical predicates
+// Returns boost::shared_ptr<Wccl::Function<Wccl::Bool> >
+// ----------------------------------------------------------------------------
predicates
[Wccl::Variables& vars]
returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > ret]
@@ -589,18 +528,28 @@ predicates
logical_predicates
[Wccl::Variables& vars]
returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > ret]
- : ret = lpred_and [vars]
- | ret = lpred_or [vars]
- | ret = lpred_nor [vars]
- | ret = boolean_v [vars]
+ : ret = lpred_and [vars]
+ | ret = lpred_or [vars]
+ | ret = lpred_nor [vars]
+ | ret = lpred_bool [vars]
+ | ret = lpred_in [vars]
+ | ret = lpred_inter [vars]
+ | ret = lpred_eq [vars]
+ | ret = lpred_regex [vars]
+// | ret = setvar_op [vars]
;
+// ----------------------------------------------------------------------------
// comma-separated predicates
logical_predicates_comma_sep
[Wccl::Variables& vars]
- returns [boost::shared_ptr<std::vector<boost::shared_ptr<Wccl::Function<Wccl::Bool> > > > ret_v]
+ returns [boost::shared_ptr<
+ std::vector<boost::shared_ptr<Wccl::Function<Wccl::Bool> > >
+ > ret_v]
{
boost::shared_ptr<Wccl::Function<Wccl::Bool> > pred;
- ret_v.reset(new std::vector<boost::shared_ptr<Wccl::Function<Wccl::Bool> > >);
+ ret_v.reset(
+ new std::vector<boost::shared_ptr<Wccl::Function<Wccl::Bool> > >
+ );
}
: pred = logical_predicates [vars] {
ret_v->push_back(pred);
@@ -622,6 +571,7 @@ lpred_and
op.reset(new Wccl::And(ret_v));
}
;
+// ----------------------------------------------------------------------------
lpred_or
[Wccl::Variables& vars]
returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > op]
@@ -634,6 +584,7 @@ lpred_or
op.reset(new Wccl::Or(ret_v));
}
;
+// ----------------------------------------------------------------------------
lpred_nor
[Wccl::Variables& vars]
returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > op]
@@ -646,180 +597,298 @@ lpred_nor
op.reset(new Wccl::Nor(ret_v));
}
;
+// ----------------------------------------------------------------------------
+lpred_bool
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > op]
+ : op = boolean [vars]
+ | op = boolean_v
+;
+// ----------------------------------------------------------------------------
+lpred_in
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > op]
+{
+ boost::shared_ptr<Wccl::Function<Wccl::TSet> > ts1, ts2;
+ boost::shared_ptr<Wccl::Function<Wccl::StrSet> > ss1, ss2;
+ boost::shared_ptr<Wccl::Function<Wccl::Position> > p1, p2;
+ boost::shared_ptr<Wccl::Function<Wccl::StrSet> > expr;
+}
+ : "in" LPAREN ss1 = string_operators [vars] COMMA
+ ss2 = string_operators [vars] RPAREN {
+ op.reset(
+// new Wccl::IsSubsetOf(*ss1.get(), *ss2.get())
+ );
+ }
+ | "in" LPAREN ts1 = sym_set_operators [vars] COMMA
+ ts2 = sym_set_operators [vars] RPAREN {
+ // op.reset(new Wccl::IsSubsetOf(*ts1.get(), *ts2.get()));
+ }
+ | "in" LPAREN p1 = position_operators [vars] COMMA
+ p2 = position_operators [vars] RPAREN {
+ // op.reset(new Wccl::IsSubsetOf(*p1.get(), *p2.get()));
+ }
+;
+// ----------------------------------------------------------------------------
+lpred_inter
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > op]
+{
+ boost::shared_ptr<Wccl::Function<Wccl::TSet> > ts1, ts2;
+ boost::shared_ptr<Wccl::Function<Wccl::StrSet> > ss1, ss2;
+ boost::shared_ptr<Wccl::Function<Wccl::Position> > p1, p2;
-// ----------------------------------------------------------------------------------
+}
+ : "inter" LPAREN ss1 = string_operators [vars] COMMA
+ ss2 = string_operators [vars] RPAREN {
+ // op.reset(new Wccl::Intersects(*ss1.get(), *ss2.get()));
+ }
+ | "inter" LPAREN ts1 = sym_set_operators [vars] COMMA
+ ts2 = sym_set_operators [vars] RPAREN {
+ // op.reset(new Wccl::Intersects(*ts1.get(), *ts2.get()));
+ }
+ | "inter" LPAREN p1 = position_operators [vars] COMMA
+ p2 = position_operators [vars] RPAREN {
+ // op.reset(new Wccl::Intersects(*p1.get(), *p2.get()));
+ }
+;
+// ----------------------------------------------------------------------------
+lpred_eq
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > op]
+{
+ boost::shared_ptr<Wccl::Function<Wccl::TSet> > ts1, ts2;
+ boost::shared_ptr<Wccl::Function<Wccl::Position> > p1, p2;
+ boost::shared_ptr<Wccl::Function<Wccl::StrSet> > ss1, ss2;
+}
+ : "equal" LPAREN ss1 = string_operators [vars] COMMA
+ ss2 = string_operators [vars] RPAREN {
+ op.reset(new Wccl::Equals<Wccl::StrSet>(ss1, ss2));
+ }
+ | "equal" LPAREN ts1 = sym_set_operators [vars] COMMA
+ ts2 = sym_set_operators [vars] RPAREN {
+ op.reset(new Wccl::Equals<Wccl::TSet>(ts1, ts2));
+ }
+ | "equal" LPAREN p1 = position_operators [vars] COMMA
+ p2 = position_operators [vars] RPAREN {
+ op.reset(new Wccl::Equals<Wccl::Position>(p1, p2));
+ }
+;
+
+// ----------------------------------------------------------------------------
+lpred_regex
+ [Wccl::Variables& vars]
+ returns [boost::shared_ptr<Wccl::Function<Wccl::Bool> > op]
+{
+ boost::shared_ptr<Wccl::Function<Wccl::StrSet> > expr;
+}
+ : "regex" LPAREN expr = string_operators [vars] COMMA reg: STRING RPAREN {
+ op.reset(new Wccl::Regex(expr, token_ref_to_ustring(reg)));
+ }
+;
+
+///////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
// ANTLR LEXER
-// ----------------------------------------------------------------------------------
+///////////////////////////////////////////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////
class ANTLRLexer extends Lexer;
options {
- k = 2;
exportVocab = ANTLRExpr;
charVocabulary = '\3'..'\377';
testLiterals = false;
+ k = 2;
}
STRING
options {
paraphrase = "a string";
}
- : '"' (~'"')* '"'
- | '\'' (~'\'')* '\''
- ;
-
-// STRING_APOS
-// options {
-// paraphrase = "a string without apostrophe";
-// }
-// : (~'"')*
-// ;
-
-// STRING_QUOT
-// options {
-// paraphrase = "a string without quotation";
-// }
-// : (~'\'')*
-// ;
-
+ : '"' (~'"')* '"'
+ | '\'' (~'\'')* '\''
+;
INT
options {
paraphrase = "Integer";
}
- : ('-'|'+')?('0'..'9')+
- ;
+ : ('-'|'+')? ('0'..'9')+
+;
QUOT_MARK
options {
paraphrase = "Quota mark";
-}
- : '\''
- ;
+}
+ : '\''
+;
APOS_MARK
options {
paraphrase = "Aposptrophe mark";
}
- : '"'
- ;
-
+ : '"'
+;
Q_MARK
options {
paraphrase = "Query mark";
}
: '?'
- ;
+;
E_MARK
options {
paraphrase = "Exclamanation mark";
}
: '!'
- ;
+;
G_MARK
options {
paraphrase = "Gravis mark";
}
: '`'
- ;
+;
+
+STR_PREFIX
+options {
+ paraphrase = "String prefix";
+}
+ : "s:"
+;
+
+TST_PREFIX
+options {
+ paraphrase = "Tag set (symbol) prefix";
+}
+ : "t:"
+;
+
+BOOL_PREFIX
+options {
+ paraphrase = "Bool prefix";
+}
+ : "b:"
+;
LBRACKET
options {
paraphrase = "'['";
}
: '['
- ;
+;
RBRACKET
options {
paraphrase = "']'";
}
: ']'
- ;
+;
LPAREN
options {
paraphrase = "'('";
}
: '('
- ;
+;
RPAREN
options {
paraphrase = "')'";
}
: ')'
- ;
+;
LCURLY
options {
paraphrase = "'{'";
}
: '{'
- ;
+;
RCURLY
options {
paraphrase = "'}'";
}
: '}'
- ;
+;
DOLLAR
options {
paraphrase = "'$'";
}
: '$'
- ;
+;
AT_MARK
options {
paraphrase = "'@'";
}
: '@'
- ;
+;
COMMA
options {
paraphrase = "','";
}
: ','
- ;
+;
SYMBOL
options {
- paraphrase = "symbol";
+ paraphrase = "Symbol";
testLiterals = true;
}
- : ( 'a'..'z' | 'A'..'Z' | '_' ) ( 'a'..'z' | 'A'..'Z' | '_' | '0'..'9' )*
- ;
+// : ('a'..'z' | 'A'..'Z' | '_') ('a'..'z' | 'A'..'Z' | '_' | '0'..'9')*
+ : ('a'..'z' | '_') ('a'..'z' | 'A'..'Z' | '_' | '0'..'9')*
+;
+
+VAR_NAME
+options {
+ paraphrase = "Variable name";
+}
+ : ('A'..'Z') ('a'..'z' | 'A'..'Z' | '_' | '0'..'9')*
+;
+
+TRUE_VALUE
+options {
+ paraphrase = "True value";
+}
+ : "True"
+;
+
+FALSE_VALUE
+options {
+ paraphrase = "False value";
+}
+ : "False"
+;
WS
- :
- ( ' '
- | '\t'
- | '\r' '\n' {newline(); }
- | '\n' {newline(); } ) { $setType(antlr::Token::SKIP); }
- ;
+ : ( ' '
+ | '\t'
+ | '\r' '\n' { newline(); }
+ | '\n' { newline(); }
+ ) { $setType(antlr::Token::SKIP); }
+;
COMMENT
options {
paraphrase = "Comment";
}
: "//" (~'\n')* '\n'{ $setType(antlr::Token::SKIP); newline(); }
- ;
+;
HASH
options {
paraphrase = "'#'";
}
: '#'
- ;
+;
DSEPARATOR
options {
paraphrase = "':-'";
}
: ":-"
- ;
+;
diff --git a/wcclparser/bool_main.cpp b/wcclparser/bool_main.cpp
new file mode 100644
index 0000000..d41f0f1
--- /dev/null
+++ b/wcclparser/bool_main.cpp
@@ -0,0 +1,77 @@
+#include <cstdlib>
+
+#include <libwccl/values/strset.h>
+#include <libwccl/parser/Parser.h>
+
+// ----------------------------------------------------------------------------
+
+/**
+ * @desc It's simple command line tester for testing predicates
+ */
+
+int main()
+{
+ std::string str_in;
+ Corpus2::Tagset tagset;
+ Parser parser(tagset);
+
+ boost::shared_ptr<const Wccl::Bool> retVal;
+ boost::shared_ptr<ANTLRParserResult<Wccl::Bool> > retOp;
+ boost::shared_ptr<Corpus2::Sentence> sentence;
+ Wccl::SentenceContext sc(sentence);
+
+ if (system("clear")) {
+ //
+ }
+
+ std::cerr << "Simple command line tester for testing bool operators"
+ << std::endl;
+
+ while (1) {
+ std::cerr << "Enter a bool operator expression: ";
+
+ getline(std::cin, str_in);
+
+ if (str_in == "clear" || str_in == "cls") {
+ if (system("clear")) {
+ //
+ }
+ }
+ else if (str_in == "exit" || str_in == "quit") {
+ break;
+ }
+ else {
+ try {
+ retOp = parser.parsePredicate(str_in);
+
+ if (retOp.get()) {
+ Wccl::FunExecContext cx(sc, retOp->variables);
+
+ if ((retVal = retOp->op->apply(cx)).get()) {
+ std::cerr << "Parsed expression: " << retVal->to_raw_string()
+ << std::endl;
+ }
+ else {
+ std::cerr << "Problem while parsing -- "
+ << "haven't got StrSet object in boost::shared_ptr!"
+ << std::endl;
+ }
+ }
+ else {
+ std::cerr << "Problem while parsing -- "
+ << "haven't got Function<Wccl::StrSet> object in "
+ << "boost::shared_ptr!" << std::endl;
+ }
+ }
+ catch (antlr::MismatchedTokenException &e) {
+ std::cerr << e.getMessage() << std::endl;
+ }
+ catch (...) {
+ std::cerr << "[N] Syntax error!" << std::endl;
+ }
+ }
+ }
+
+ return 0;
+}
+
diff --git a/wcclparser/strop_main.cpp b/wcclparser/strop_main.cpp
index 4f4d485..255e81d 100644
--- a/wcclparser/strop_main.cpp
+++ b/wcclparser/strop_main.cpp
@@ -1,67 +1,77 @@
#include <cstdlib>
+#include <boost/shared_ptr.hpp>
#include <libwccl/values/strset.h>
#include <libwccl/parser/Parser.h>
+#include <libwccl/parser/ANTLRParserResult.h>
// ----------------------------------------------------------------------------
/**
* @desc It's simple command line tester for testing string operators
*/
-
int main()
{
std::string str_in;
Corpus2::Tagset tagset;
Parser parser(tagset);
- boost::shared_ptr<Wccl::StrSet> retStr;
- boost::shared_ptr<Wccl::Function<Wccl::StrSet> > retOp;
+ boost::shared_ptr<const Wccl::StrSet> retVal;
+ boost::shared_ptr<ANTLRParserResult<Wccl::StrSet> > retOp;
boost::shared_ptr<Corpus2::Sentence> sentence;
Wccl::SentenceContext sc(sentence);
- while (1) {
- std::cerr << "Enter a string operator expression: ";
+ if (system("clear")) {
+ //
+ }
+
+ std::cerr << "Simple command line tester for testing string operators"
+ << std::endl;
- getline(std::cin, str_in);
+ while (1) {
+ std::cerr << "Enter a string operator expression: ";
- if (str_in == "clear" || str_in == "cls") {
- if (system("clear")) {
- //
- }
- }
+ getline(std::cin, str_in);
+
+ if (str_in == "clear" || str_in == "cls") {
+ if (system("clear")) {
+ //
+ }
+ }
else if (str_in == "exit" || str_in == "quit") {
break;
}
- else {
+ else {
try {
- retOp = parser.parseStringOperator(str_in);
+ retOp = parser.parseStringOperator(str_in);
if (retOp.get()) {
- if ((retStr = retOp->apply(sc)).get()) {
- std::cerr << "Parsed expression: " << retStr->to_raw_string() << std::endl;
+ Wccl::FunExecContext cx(sc, retOp->variables);
+
+ if ((retVal = retOp->op->apply(cx)).get()) {
+ std::cerr << "Parsed expression: " << retVal->to_raw_string()
+ << std::endl;
}
else {
- std::cerr << "Problem while parsing -- haven't got StrSet object in boost::shared_ptr!" << std::endl;
+ std::cerr << "Problem while parsing -- "
+ << "haven't got StrSet object in boost::shared_ptr!"
+ << std::endl;
}
}
else {
- std::cerr << "Problem while parsing -- haven't got Function<Wccl::StrSet> object in boost::shared_ptr!" << std::endl;
+ std::cerr << "Problem while parsing -- "
+ << "haven't got Function<Wccl::StrSet> object in "
+ << "boost::shared_ptr!" << std::endl;
}
}
catch (antlr::MismatchedTokenException &e) {
- std::cerr << "Mismatch token exception!" << std::endl;
+ std::cerr << e.getMessage() << std::endl;
}
- /*
- catch (antlr::TokenStreamRecognitionException &e) {
- std::cerr << "[2] Syntax error!" << std::endl;
- }
- */
catch (...) {
std::cerr << "[N] Syntax error!" << std::endl;
}
- }
- }
+ }
+ }
return 0;
}
diff --git a/wcclparser/tagset_main.cpp b/wcclparser/tagset_main.cpp
new file mode 100644
index 0000000..1dd25c6
--- /dev/null
+++ b/wcclparser/tagset_main.cpp
@@ -0,0 +1,78 @@
+#include <cstdlib>
+
+#include <libwccl/values/tset.h>
+
+#include <libwccl/parser/Parser.h>
+#include <libwccl/parser/ANTLRParserResult.h>
+
+// ----------------------------------------------------------------------------
+
+/**
+ * @desc It's simple command line tester for testing tagset operators
+ */
+int main()
+{
+ std::string str_in;
+ Corpus2::Tagset tagset;
+ Parser parser(tagset);
+
+ boost::shared_ptr<const Wccl::TSet> retVal;
+ boost::shared_ptr<ANTLRParserResult<Wccl::TSet> > retOp;
+ boost::shared_ptr<Corpus2::Sentence> sentence;
+ Wccl::SentenceContext sc(sentence);
+
+ if (system("clear")) {
+ //
+ }
+
+ std::cerr << "Simple command line tester for testing bool operators"
+ << std::endl;
+
+ while (1) {
+ std::cerr << "Enter a bool operator expression: ";
+
+ getline(std::cin, str_in);
+
+ if (str_in == "clear" || str_in == "cls") {
+ if (system("clear")) {
+ //
+ }
+ }
+ else if (str_in == "exit" || str_in == "quit") {
+ break;
+ }
+ else {
+ try {
+ retOp = parser.parseSymSetOperator(str_in);
+
+ if (retOp.get()) {
+ Wccl::FunExecContext cx(sc, retOp->variables);
+
+ if ((retVal = retOp->op->apply(cx)).get()) {
+ std::cerr << "Parsed expression: " << retVal->to_raw_string()
+ << std::endl;
+ }
+ else {
+ std::cerr << "Problem while parsing -- "
+ << "haven't got StrSet object in boost::shared_ptr!"
+ << std::endl;
+ }
+ }
+ else {
+ std::cerr << "Problem while parsing -- "
+ << "haven't got Function<Wccl::StrSet> object in "
+ << "boost::shared_ptr!" << std::endl;
+ }
+ }
+ catch (antlr::MismatchedTokenException &e) {
+ std::cerr << e.getMessage() << std::endl;
+ }
+ catch (...) {
+ std::cerr << "[N] Syntax error!" << std::endl;
+ }
+ }
+ }
+
+ return 0;
+}
+
--
GitLab