From 8aa9c414cb19558650c8deaa91daf47471bc639f Mon Sep 17 00:00:00 2001
From: Adam Radziszewski <adam.radziszewski@pwr.wroc.pl>
Date: Wed, 13 Jul 2011 14:38:41 +0200
Subject: [PATCH] tagging: token as tag projection
---
libcorpus2/tagging.cpp | 16 ++++++++++++++--
libcorpus2/tagging.h | 7 +++++++
swig/tagging.i | 6 +++++-
3 files changed, 26 insertions(+), 3 deletions(-)
diff --git a/libcorpus2/tagging.cpp b/libcorpus2/tagging.cpp
index b98aee1..b1f4587 100644
--- a/libcorpus2/tagging.cpp
+++ b/libcorpus2/tagging.cpp
@@ -15,9 +15,10 @@ or FITNESS FOR A PARTICULAR PURPOSE.
*/
#include <libcorpus2/tagging.h>
-#include <libcorpus2/tagsetmanager.h>
+#include <libcorpus2/lexeme.h>
-// #include <libpwrutils/foreach.h>
+
+#include <libpwrutils/foreach.h>
namespace Corpus2 {
@@ -40,5 +41,16 @@ Tag get_attribute_mask(const Tagset& tagset, std::string attr_name)
}
}
+Tag mask_token(const Token& token, const Tag& mask, bool disamb_only)
+{
+ Tag t;
+ foreach (const Corpus2::Lexeme& lexeme, token.lexemes()) {
+ if(lexeme.is_disamb() || !disamb_only) {
+ t.combine_with(lexeme.tag().get_masked(mask));
+ }
+ }
+ return t;
+}
+
} /* end ns Corpus2 */
diff --git a/libcorpus2/tagging.h b/libcorpus2/tagging.h
index b767a85..42e4dd2 100644
--- a/libcorpus2/tagging.h
+++ b/libcorpus2/tagging.h
@@ -18,6 +18,7 @@ or FITNESS FOR A PARTICULAR PURPOSE.
#define LIBCORPUS2_TAGGING_H
#include <libcorpus2/tag.h>
+#include <libcorpus2/token.h>
#include <libcorpus2/tagset.h>
namespace Corpus2 {
@@ -35,6 +36,12 @@ namespace Corpus2 {
Tag get_attribute_mask(const Tagset& tagset,
const std::string attr_name);
+/**
+ * Projects the token onto the mask. Depending on disamb_only, will consider
+ * only disamb or all lexemes.
+ */
+Tag mask_token(const Token& token, const Tag& mask, bool disamb_only);
+
} /* end ns Corpus2 */
#endif // LIBCORPUS2_TAGGING_H
diff --git a/swig/tagging.i b/swig/tagging.i
index 8d2301a..21c79f1 100644
--- a/swig/tagging.i
+++ b/swig/tagging.i
@@ -10,11 +10,15 @@
%include "tag.i"
%include "tagset.i"
+%include "token.i"
namespace Corpus2 {
Tag get_attribute_mask(const Tagset& tagset,
- const std::string attr_name);
+ const std::string attr_name);
+
+Tag mask_token(const Token& token, const Tag& mask, bool disamb_only);
+
}
using namespace std;
--
GitLab