From b000d45bd7c08d268495d3974cb3dd8fbdebac94 Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Mon, 11 Oct 2010 11:41:24 +0200
Subject: [PATCH] preliminary disamb_sh support in sces reader, bumps version
to 0.0.2
---
libcorpus2/CMakeLists.txt | 2 +-
libcorpus2/io/xcesreader.cpp | 27 +++++++++++++++++++--------
libcorpus2/io/xcesreader.h | 2 +-
3 files changed, 21 insertions(+), 10 deletions(-)
diff --git a/libcorpus2/CMakeLists.txt b/libcorpus2/CMakeLists.txt
index 424099b..3cf1325 100644
--- a/libcorpus2/CMakeLists.txt
+++ b/libcorpus2/CMakeLists.txt
@@ -3,7 +3,7 @@ PROJECT(corpus2)
set(corpus2_ver_major "0")
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/io/xcesreader.cpp b/libcorpus2/io/xcesreader.cpp
index 03e8f8c..af32af0 100644
--- a/libcorpus2/io/xcesreader.cpp
+++ b/libcorpus2/io/xcesreader.cpp
@@ -9,7 +9,7 @@ class XcesReaderImpl : public BasicSaxParser
{
public:
XcesReaderImpl(const Tagset& tagset, std::deque<Chunk*>& obuf,
- bool disamb_only);
+ bool disamb_only, bool disamb_sh);
~XcesReaderImpl();
@@ -37,12 +37,14 @@ protected:
std::deque<Chunk*>& obuf_;
bool disamb_only_;
+
+ bool disamb_sh_;
};
XcesReader::XcesReader(const Tagset& tagset, std::istream& is,
- bool disamb_only)
+ bool disamb_only, bool disamb_sh)
: BufferedChunkReader(tagset), is_(is)
- , impl_(new XcesReaderImpl(tagset, chunk_buf_, disamb_only))
+ , impl_(new XcesReaderImpl(tagset, chunk_buf_, disamb_only, disamb_sh))
{
}
@@ -64,11 +66,11 @@ void XcesReader::ensure_more()
}
XcesReaderImpl::XcesReaderImpl(const Tagset& tagset,
- std::deque<Chunk*>& obuf, bool disamb_only)
+ std::deque<Chunk*>& obuf, bool disamb_only, bool disamb_sh)
: BasicSaxParser()
, tagset_(tagset), state_(XS_NONE), wa_(PwrNlp::Whitespace::Newline)
, sbuf_(), tok_(NULL), sent_(NULL), chunk_(NULL), obuf_(obuf)
- , disamb_only_(disamb_only)
+ , disamb_only_(disamb_only), disamb_sh_(disamb_sh)
{
}
@@ -119,9 +121,18 @@ void XcesReaderImpl::on_start_element(const Glib::ustring &name,
} else if (state_ == XS_TOK && name == "lex") {
assert(tok_ != NULL);
bool is_disamb = false;
- foreach (const Attribute& a, attributes) {
- if (a.name == "disamb" && a.value == "1") {
- is_disamb = true;
+ if (!disamb_sh_) {
+ foreach (const Attribute& a, attributes) {
+ if (a.name == "disamb" && a.value == "1") {
+ is_disamb = true;
+ }
+ }
+ } else {
+ is_disamb = true;
+ foreach (const Attribute& a, attributes) {
+ if (a.name == "disamb_sh" && a.value == "0") {
+ is_disamb = false;
+ }
}
}
if (!disamb_only_ || is_disamb) {
diff --git a/libcorpus2/io/xcesreader.h b/libcorpus2/io/xcesreader.h
index 2af10d7..457a8aa 100644
--- a/libcorpus2/io/xcesreader.h
+++ b/libcorpus2/io/xcesreader.h
@@ -15,7 +15,7 @@ class XcesReader : public BufferedChunkReader
{
public:
XcesReader(const Tagset& tagset, std::istream& is,
- bool disamb_only = false);
+ bool disamb_only = false, bool disamb_sh = false);
~XcesReader();
--
GitLab