From 071f2b1b234eda8ce153913940dae348696f8b91 Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Tue, 19 Apr 2011 15:37:21 +0200
Subject: [PATCH] fix xces/ccl ,disamb_only processing
---
libcorpus2/io/xmlreader.cpp | 6 +++++-
libcorpus2/io/xmlreader.h | 2 ++
2 files changed, 7 insertions(+), 1 deletion(-)
diff --git a/libcorpus2/io/xmlreader.cpp b/libcorpus2/io/xmlreader.cpp
index 85ba6e9..7d5ed6a 100644
--- a/libcorpus2/io/xmlreader.cpp
+++ b/libcorpus2/io/xmlreader.cpp
@@ -77,6 +77,8 @@ void XmlReader::on_start_element(const Glib::ustring &name,
state_ = STATE_TAG;
grab_characters_ = true;
clear_buf();
+ } else if (state_ == STATE_LEX_SKIP && name == "lex" || name == "base" || name == "ctag") {
+ //nop
} else if (name == "ns") {
wa_ = PwrNlp::Whitespace::None;
} else if (state_ == STATE_NONE && name == "tok") {
@@ -175,6 +177,8 @@ void XmlReader::start_lexeme(const AttributeList &attributes)
tok_->add_lexeme(Lexeme());
tok_->lexemes().back().set_disamb(is_disamb);
state_ = STATE_LEX;
+ } else {
+ state_ = STATE_LEX_SKIP;
}
}
@@ -225,7 +229,7 @@ void XmlReader::on_end_element(const Glib::ustring &name)
tok_->lexemes().back().set_tag(tag);
grab_characters_ = false;
state_ = STATE_LEX;
- } else if (state_ == STATE_LEX && name == "lex") {
+ } else if ((state_ == STATE_LEX || state_ == STATE_LEX_SKIP) && name == "lex") {
state_ = STATE_TOK;
} else if (state_ == STATE_TOK && name == "tok") {
finish_token();
diff --git a/libcorpus2/io/xmlreader.h b/libcorpus2/io/xmlreader.h
index 7b87571..d9234d4 100644
--- a/libcorpus2/io/xmlreader.h
+++ b/libcorpus2/io/xmlreader.h
@@ -84,6 +84,8 @@ protected:
static const int STATE_LEX = 5;
static const int STATE_LEMMA = 6;
static const int STATE_TAG = 7;
+ static const int STATE_LEX_SKIP = 8;
+
/// The state of the parser
int state_;
--
GitLab