From 54f4cd4241bcddff3290284df57fbc5de1d75f6b Mon Sep 17 00:00:00 2001
From: Adam Radziszewski <adam.radziszewski@pwr.wroc.pl>
Date: Tue, 5 Jun 2012 15:23:12 +0200
Subject: [PATCH] new token ann checking preds: isannbeg, isannend, isannhead;
advance ver
---
CMakeLists.txt | 4 ++--
.../functions/bool/predicates/anntoken.cpp | 24 ++++++++++++++++---
.../ops/functions/bool/predicates/anntoken.h | 20 +++++++++++++---
libwccl/parser/grammar.g | 8 +++----
4 files changed, 44 insertions(+), 12 deletions(-)
diff --git a/CMakeLists.txt b/CMakeLists.txt
index aedc2ca..258b43d 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -2,8 +2,8 @@ PROJECT(WCCL)
cmake_minimum_required(VERSION 2.8.0)
set(wccl_ver_major "0")
-set(wccl_ver_minor "1")
-set(wccl_ver_patch "6")
+set(wccl_ver_minor "2")
+set(wccl_ver_patch "0")
set(LIBWCCL_VERSION "${wccl_ver_major}.${wccl_ver_minor}.${wccl_ver_patch}")
diff --git a/libwccl/ops/functions/bool/predicates/anntoken.cpp b/libwccl/ops/functions/bool/predicates/anntoken.cpp
index 9f2adc9..83cc101 100644
--- a/libwccl/ops/functions/bool/predicates/anntoken.cpp
+++ b/libwccl/ops/functions/bool/predicates/anntoken.cpp
@@ -42,15 +42,33 @@ AnnToken::BaseRetValPtr AnnToken::apply_internal(const FunExecContext& context)
// channel exists, position in range, check if any annot at pos
int abs_idx = sc.get_abs_position(*pos);
- if (as->get_channel(chan_name_).get_segment_at(abs_idx) == 0) {
+ const Corpus2::AnnotationChannel& chan = as->get_channel(chan_name_);
+
+ if (chan.get_segment_at(abs_idx) == 0) {
return Predicate::False(context);
}
else {
- return Predicate::True(context);
+ if (tok_constr_ == O_ANY) {
+ return Predicate::True(context);
+ }
+ if (tok_constr_ == O_HEAD) {
+ return chan.is_head_at(abs_idx) ?
+ Predicate::True(context) : Predicate::False(context);
+ }
+ if (tok_constr_ == O_FIRST) {
+ return (chan.get_segment_at(abs_idx) !=
+ chan.get_segment_at(abs_idx - 1)) ?
+ Predicate::True(context) : Predicate::False(context);
+ }
+ if (tok_constr_ == O_LAST) {
+ return (chan.get_segment_at(abs_idx) !=
+ chan.get_segment_at(abs_idx + 1)) ?
+ Predicate::True(context) : Predicate::False(context);
+ }
}
}
-std::string AnnToken::to_string(const Corpus2::Tagset& tagset) const
+std::string AnnToken::to_string(const Corpus2::Tagset& /* tagset */) const
{
std::ostringstream ostream;
ostream << raw_name() << "(" << *pos_expr_ << ", "
diff --git a/libwccl/ops/functions/bool/predicates/anntoken.h b/libwccl/ops/functions/bool/predicates/anntoken.h
index b162764..179eda4 100644
--- a/libwccl/ops/functions/bool/predicates/anntoken.h
+++ b/libwccl/ops/functions/bool/predicates/anntoken.h
@@ -24,17 +24,30 @@ namespace Wccl {
/**
* An annotation-checking predicate: checks whether the given position belongs
- * to any annotation in the given channel.
+ * to any annotation in the given channel. If ann constraint provided, will
+ * return True only if the additional constraint holds: token must be
+* annotation's head, must be annotation-first or annotation-last.
*/
class AnnToken : public Predicate
{
public:
typedef boost::shared_ptr<Function<Position> > PosFunctionPtr;
+
+ enum TokenConstraint
+ {
+ O_ANY = 0, // no constraints
+ O_HEAD = 1, // token must be ann head
+ O_FIRST = 2, // must be ann-first
+ O_LAST = 3, // must be ann-last
+ };
+
AnnToken(
const PosFunctionPtr& pos_expr,
- const std::string& annotation_name)
+ const std::string& annotation_name,
+ const TokenConstraint tok_constr = O_ANY)
: pos_expr_(pos_expr),
- chan_name_(annotation_name)
+ chan_name_(annotation_name),
+ tok_constr_(tok_constr)
{
BOOST_ASSERT(pos_expr_);
BOOST_ASSERT(!annotation_name.empty());
@@ -55,6 +68,7 @@ public:
protected:
const PosFunctionPtr pos_expr_;
const std::string chan_name_;
+ const TokenConstraint tok_constr_;
/**
* Returns whether the given position belongs to an annotation in
diff --git a/libwccl/parser/grammar.g b/libwccl/parser/grammar.g
index 2e146ab..167e390 100644
--- a/libwccl/parser/grammar.g
+++ b/libwccl/parser/grammar.g
@@ -1786,22 +1786,22 @@ bool_phrase_annotation
: "isannpart" LPAREN
lpos = position_operator [scope] COMMA n1: STRING
RPAREN {
- ret.reset(new AnnToken(lpos, token_ref_to_std_string(n1)));
+ ret.reset(new AnnToken(lpos, token_ref_to_std_string(n1), AnnToken::O_ANY));
}
| "isannbeg" LPAREN
lpos = position_operator [scope] COMMA n2: STRING
RPAREN {
- // TODO
+ ret.reset(new AnnToken(lpos, token_ref_to_std_string(n2), AnnToken::O_FIRST));
}
| "isannend" LPAREN
lpos = position_operator [scope] COMMA n3: STRING
RPAREN {
- // TODO
+ ret.reset(new AnnToken(lpos, token_ref_to_std_string(n3), AnnToken::O_LAST));
}
| "isannhead" LPAREN
lpos = position_operator [scope] COMMA n4: STRING
RPAREN {
- // TODO
+ ret.reset(new AnnToken(lpos, token_ref_to_std_string(n4), AnnToken::O_HEAD));
}
| "isannwhole" LPAREN
lpos = position_operator [scope] COMMA
--
GitLab