From 5db3ef6d6ab8166d058af52753dabece31a2ff94 Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Fri, 1 Oct 2010 12:16:38 +0200
Subject: [PATCH] Fix disamb ordering in Lexeme op< add disamb-ignoring
equality use both of these in Token's remove duplicate lexemes to
mEdukacjaWeb/bilanse.do?event=switchBilanse&bilanse=prognozowaneake disamb
take priority in case of a disamb/ndisamb dupe
---
libcorpus2/lexeme.cpp | 7 ++++++-
libcorpus2/lexeme.h | 8 ++++++++
libcorpus2/token.cpp | 2 +-
3 files changed, 15 insertions(+), 2 deletions(-)
diff --git a/libcorpus2/lexeme.cpp b/libcorpus2/lexeme.cpp
index 9714860..cd53613 100644
--- a/libcorpus2/lexeme.cpp
+++ b/libcorpus2/lexeme.cpp
@@ -28,7 +28,7 @@ bool Lexeme::operator<(const Lexeme& other) const
|| (lemma_ == other.lemma_
&& (tag_ < other.tag_
|| (tag_ == other.tag_
- && disamb_ < other.disamb_)));
+ && disamb_ && !other.disamb_)));
}
bool Lexeme::operator==(const Lexeme& other) const
@@ -37,4 +37,9 @@ bool Lexeme::operator==(const Lexeme& other) const
disamb_ == other.disamb_;
}
+bool Lexeme::DisamblessComparator::operator()(const Lexeme& l, const Lexeme& other) const
+{
+ return l.lemma_ == other.lemma_ && l.tag_ == other.tag_;
+}
+
} /* end ns Corpus2 */
diff --git a/libcorpus2/lexeme.h b/libcorpus2/lexeme.h
index 1e70d17..0a508b1 100644
--- a/libcorpus2/lexeme.h
+++ b/libcorpus2/lexeme.h
@@ -88,6 +88,14 @@ public:
*/
bool operator==(const Lexeme& other) const;
+ /**
+ * Disamb-ignoring lexeme comparison
+ */
+ struct DisamblessComparator
+ {
+ bool operator()(const Lexeme& l1, const Lexeme& l2) const;
+ };
+
private:
/// The lemma -- basic form
//boost::flyweight<UnicodeString> lemma_;
diff --git a/libcorpus2/token.cpp b/libcorpus2/token.cpp
index f4317c5..ac86d5c 100644
--- a/libcorpus2/token.cpp
+++ b/libcorpus2/token.cpp
@@ -84,7 +84,7 @@ bool Token::remove_duplicate_lexemes()
{
size_t old_size = lexemes_.size();
std::sort(lexemes_.begin(), lexemes_.end());
- lexemes_.erase(std::unique(lexemes_.begin(), lexemes_.end()),
+ lexemes_.erase(std::unique(lexemes_.begin(), lexemes_.end(), Lexeme::DisamblessComparator()),
lexemes_.end());
return old_size != lexemes_.size();
}
--
GitLab