From d35e12c3b1a195d28ee710624f72ae5650b68b25 Mon Sep 17 00:00:00 2001
From: Bartosz Broda <bartosz.broda@gmail.com>
Date: Tue, 14 Jun 2011 09:06:52 +0200
Subject: [PATCH] add searching for head of an mwe
---
libmwereader/mwe.cpp | 23 ++++++++++++++++++++---
libmwereader/mwe.h | 2 +-
libmwereader/mwereader.cpp | 3 ++-
libmwereader/test_mwe.xml | 2 +-
4 files changed, 24 insertions(+), 6 deletions(-)
diff --git a/libmwereader/mwe.cpp b/libmwereader/mwe.cpp
index cc1f537..5799d2c 100644
--- a/libmwereader/mwe.cpp
+++ b/libmwereader/mwe.cpp
@@ -24,7 +24,7 @@ LexicalUnit::LexicalUnit(const std::string &base,
}
bool LexicalUnit::IsHere(const Wccl::SentenceContext &sc,
- std::set<size_t> &out_position) const
+ std::set<size_t> &out_position, int &head_pos) const
{
// set variables
for(variables_map::const_iterator ivars = variables_.begin();
@@ -37,6 +37,10 @@ bool LexicalUnit::IsHere(const Wccl::SentenceContext &sc,
if(pResult->get_value() == false)
return false;
+ bool found_head = false;
+
+ Wccl::SentenceContext sc2(sc.get_sentence_ptr());
+
// fill up positions
foreach(const std::string&varname, condition_->valid_variable_names()){
if(boost::algorithm::starts_with(varname, "Pos")){
@@ -46,11 +50,24 @@ bool LexicalUnit::IsHere(const Wccl::SentenceContext &sc,
errmsg += " Offending unit: " + base_;
throw Wccl::WcclError(errmsg);
}
- out_position.insert( sc.get_abs_position(pos) );
+ int abs_pos = sc.get_abs_position(pos);
+ out_position.insert( abs_pos );
+ if(!found_head){
+ sc2.set_position(abs_pos);
+ if(head_cond_->apply(sc2)->get_value()){
+ head_pos = abs_pos;
+ found_head = true;
+ }
+ }
}
}
- // TODO head position
+ if(!found_head){
+ std::string errmsg("MWE found, but no head for it.");
+ errmsg += " Offending unit: " + base_;
+ throw Wccl::WcclError(errmsg);
+ }
+
return true;
}
diff --git a/libmwereader/mwe.h b/libmwereader/mwe.h
index 77a38fe..439323f 100644
--- a/libmwereader/mwe.h
+++ b/libmwereader/mwe.h
@@ -35,7 +35,7 @@ public:
* \returns true if this lexical unit was found here
*/
virtual bool IsHere(const Wccl::SentenceContext& sc,
- std::set<size_t> &out_positions) const;
+ std::set<size_t> &out_positions, int &head_pos) const;
const std::string & get_base() const{ return base_;}
const variables_map & get_variables() const{ return variables_;}
diff --git a/libmwereader/mwereader.cpp b/libmwereader/mwereader.cpp
index 26e4c6a..e6328be 100644
--- a/libmwereader/mwereader.cpp
+++ b/libmwereader/mwereader.cpp
@@ -46,7 +46,8 @@ bool MWEReader::registered = TokenReader::register_path_reader<MWEReader>(
std::cout << "# ";
foreach(LexicalUnit::Ptr pLU, potential){
std::set<size_t> positions;
- bool is_here = pLU->IsHere(sc, positions);
+ int head;
+ bool is_here = pLU->IsHere(sc, positions, head);
if(is_here)
std::cout << "** " << pLU->get_base() << "** ";
}
diff --git a/libmwereader/test_mwe.xml b/libmwereader/test_mwe.xml
index 43a302e..020213f 100644
--- a/libmwereader/test_mwe.xml
+++ b/libmwereader/test_mwe.xml
@@ -17,7 +17,7 @@
<MWE base="Jan Paweł">
<var name="Subst1">jan</var>
<var name="Subst2">paweł</var>
- <head>inter(cas[0], {nom})</head>
+ <head>inter(base[0], "jan")</head>
</MWE>
<MWE base="waga netto">
<var name="Subst1">waga</var>
--
GitLab