From a2a05ba0a9c3c5852f9f1af7c28f1f55ee4cb8ff Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Tue, 30 Nov 2010 14:29:28 +0100
Subject: [PATCH] add Tagset attribute mask to index/name mappingand related
functions. Bump version to 1.0.2
---
libcorpus2/CMakeLists.txt | 2 +-
libcorpus2/tagset.cpp | 22 ++++++++++++++++++++++
libcorpus2/tagset.h | 13 ++++++++++++-
libcorpus2/tagsetparser.cpp | 2 ++
tests/tag_split.cpp | 14 ++++++++++++++
5 files changed, 51 insertions(+), 2 deletions(-)
diff --git a/libcorpus2/CMakeLists.txt b/libcorpus2/CMakeLists.txt
index bd66f8d..1b19902 100644
--- a/libcorpus2/CMakeLists.txt
+++ b/libcorpus2/CMakeLists.txt
@@ -3,7 +3,7 @@ PROJECT(corpus2)
set(corpus2_ver_major "1")
set(corpus2_ver_minor "0")
-set(corpus2_ver_patch "1")
+set(corpus2_ver_patch "2")
if(NOT LIBCORPUS2_SRC_DATA_DIR)
diff --git a/libcorpus2/tagset.cpp b/libcorpus2/tagset.cpp
index 8f3a4f6..475bf10 100644
--- a/libcorpus2/tagset.cpp
+++ b/libcorpus2/tagset.cpp
@@ -535,11 +535,33 @@ idx_t Tagset::get_attribute_index(const string_range& a) const
return attribute_dict_.get_id(a);
}
+idx_t Tagset::get_attribute_index(mask_t a) const
+{
+ std::map<mask_t, idx_t>::const_iterator ci;
+ ci = attribute_mask_to_index_.find(a);
+ if (ci == attribute_mask_to_index_.end()) {
+ return -1;
+ } else {
+ return ci->second;
+ }
+}
+
const std::string& Tagset::get_attribute_name(idx_t a) const
{
return attribute_dict_.get_string(a);
}
+const std::string& Tagset::get_attribute_name(mask_t a) const
+{
+ static std::string nullstr;
+ idx_t index = get_attribute_index(a);
+ if (index < 0 || index > attribute_count()) {
+ return nullstr;
+ } else {
+ return attribute_dict_.get_string(index);
+ }
+}
+
const std::vector<mask_t>& Tagset::get_attribute_values(idx_t a) const
{
static std::vector<mask_t> null_vec;
diff --git a/libcorpus2/tagset.h b/libcorpus2/tagset.h
index f413d56..4b199b6 100644
--- a/libcorpus2/tagset.h
+++ b/libcorpus2/tagset.h
@@ -375,9 +375,17 @@ public:
/// @returns -1 on invalid name
idx_t get_attribute_index(const string_range& a) const;
+ /// Attribute mask -> index mapping
+ /// @returns -1 on invalid mask
+ idx_t get_attribute_index(mask_t a) const;
+
/// Attribute index -> name
/// @returns empty string on invalid index
- const std::string& get_attribute_name(idx_t pos) const;
+ const std::string& get_attribute_name(idx_t a) const;
+
+ /// Attribute mask -> name
+ /// @returns empty string on invalid mask
+ const std::string& get_attribute_name(mask_t a) const;
/// Value mask -> attribute index mapping.
/// if the value mask contains values from more than one attribute,
@@ -577,6 +585,9 @@ private:
/// Attribute index to combined value mask
std::vector<mask_t> attribute_masks_;
+ /// Attribute combined mask to attribute name
+ std::map<mask_t, idx_t> attribute_mask_to_index_;
+
/// reverse mapping, from a value mask to the respective attribute
/// index (values are assumed to be unique and not shared between
/// attributes)
diff --git a/libcorpus2/tagsetparser.cpp b/libcorpus2/tagsetparser.cpp
index 4a4307b..93da97c 100644
--- a/libcorpus2/tagsetparser.cpp
+++ b/libcorpus2/tagsetparser.cpp
@@ -117,6 +117,8 @@ Tagset TagsetParser::load_ini(std::istream &is)
current_value <<= 1;
}
tagset.attribute_masks_.push_back(attribute_mask);
+ tagset.attribute_mask_to_index_.insert(std::make_pair(
+ attribute_mask, current_attribute_index));
++current_attribute_index;
}
tagset.attribute_dict_.load_sorted_data(vec);
diff --git a/tests/tag_split.cpp b/tests/tag_split.cpp
index f474931..4258b7b 100644
--- a/tests/tag_split.cpp
+++ b/tests/tag_split.cpp
@@ -245,6 +245,16 @@ BOOST_FIXTURE_TEST_CASE( symbols, F )
BOOST_CHECK_THROW(tagset->parse_symbol("asdf"), Corpus2::TagParseError);
}
+BOOST_FIXTURE_TEST_CASE(attribute_mask_to_name, F)
+{
+ foreach (Corpus2::mask_t a, tagset->all_attribute_masks()) {
+ std::string aname = tagset->get_attribute_name(a);
+ BOOST_CHECK(!aname.empty());
+ Corpus2::mask_t aa = tagset->get_attribute_mask(aname);
+ BOOST_CHECK_EQUAL(a, aa);
+ }
+}
+
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_CASE(bs_split)
@@ -257,3 +267,7 @@ BOOST_AUTO_TEST_CASE(bs_split)
}
BOOST_CHECK_EQUAL(x, y);
}
+
+
+
+
--
GitLab