From 44625c67025676b1d1d73b0490053aef930a9db9 Mon Sep 17 00:00:00 2001
From: Bartosz Broda <bartosz.broda@gmail.com>
Date: Fri, 10 Jun 2011 21:00:23 +0200
Subject: [PATCH] some work on mwe parsing
---
libmwereader/mweparser.cpp | 12 ++++++++++--
libmwereader/mweparser.h | 18 +++++++++++++++++-
libmwereader/mwereader.cpp | 2 +-
libmwereader/mwereader.h | 2 +-
libmwereader/mwertest.cpp | 6 ++++--
5 files changed, 33 insertions(+), 7 deletions(-)
diff --git a/libmwereader/mweparser.cpp b/libmwereader/mweparser.cpp
index 6cedc52..add215e 100644
--- a/libmwereader/mweparser.cpp
+++ b/libmwereader/mweparser.cpp
@@ -1,6 +1,7 @@
#include "mweparser.h"
#include <libpwrutils/foreach.h>
+#include <libcorpus2/tagsetmanager.h>
#include <libxml++/libxml++.h>
#include <libxml2/libxml/parser.h>
@@ -9,8 +10,14 @@
namespace Corpus2 {
- MWEParser::MWEParser()
- : BasicSaxParser(), state_(NONE)
+ MWEBuilder::MWEBuilder(const Tagset& tagset)
+ : tagset_(tagset)
+ {
+
+ }
+
+ MWEParser::MWEParser(MWEIndex &index)
+ : BasicSaxParser(), state_(NONE), mwe_index_(index)
{
}
@@ -47,6 +54,7 @@ namespace Corpus2 {
if(state_ == NONE && name == "units_description"){
tagset_ = get_attribute(attributes, "tagset");
+ mwe_builder_ = boost::shared_ptr<MWEBuilder>(new MWEBuilder(Corpus2::get_named_tagset(tagset_)));
state_ = UNITSDESC;
} else if(state_ == UNITSDESC && name == "mwegroup"){
parse_mwegroup_attributes(attributes);
diff --git a/libmwereader/mweparser.h b/libmwereader/mweparser.h
index 81313e5..85cde5f 100644
--- a/libmwereader/mweparser.h
+++ b/libmwereader/mweparser.h
@@ -3,18 +3,31 @@
#include <libcorpus2/io/reader.h>
#include <libcorpus2/io/sax.h>
+#include <boost/unordered_map.hpp>
+
+#include "mwe.h"
namespace Corpus2 {
class MWEBuilder
{
+public:
+ MWEBuilder(const Tagset& tagset);
+ typedef boost::unordered_map<std::string, std::string> value_type;
+
+private:
+ const Tagset& tagset_;
+ /// str -> ptr to ccl operator
+ value_type main_conditions_;
+ /// str -> ptr to ccl operator
+ value_type head_conditions_;
};
class MWEParser : public BasicSaxParser
{
public:
- MWEParser();
+ MWEParser(MWEIndex &index);
protected:
typedef std::map<std::string, std::string> str_map;
@@ -56,6 +69,9 @@ protected:
std::string group_type_;
std::string group_class_;
std::string head_cond_;
+
+ MWEIndex &mwe_index_;
+ boost::shared_ptr<MWEBuilder> mwe_builder_;
};
} // ns Corpus2
diff --git a/libmwereader/mwereader.cpp b/libmwereader/mwereader.cpp
index b8e0909..d7b38c7 100644
--- a/libmwereader/mwereader.cpp
+++ b/libmwereader/mwereader.cpp
@@ -73,7 +73,7 @@ bool MWEReader::registered = TokenReader::register_path_reader<MWEReader>(
void MWEReader::load_mwes(const std::string &filename)
{
- MWEParser parser;
+ MWEParser parser(mwe_index_);
parser.parse_file(filename);
}
diff --git a/libmwereader/mwereader.h b/libmwereader/mwereader.h
index 7b8b4d9..4c65c2e 100644
--- a/libmwereader/mwereader.h
+++ b/libmwereader/mwereader.h
@@ -49,7 +49,7 @@ public:
private:
void load_mwes(const std::string& filename);
- //MWEIndex mwe_index_;
+ MWEIndex mwe_index_;
/// ptr to inner reader doing the real work of reading a corpus
TokenReaderPtr inner_reader_;
/// path for inner reader
diff --git a/libmwereader/mwertest.cpp b/libmwereader/mwertest.cpp
index cb78a83..c5b9ceb 100644
--- a/libmwereader/mwertest.cpp
+++ b/libmwereader/mwertest.cpp
@@ -5,8 +5,10 @@
int main(int ac, char**av)
{
using namespace Corpus2;
- std::cout << "Starting tests" << std::endl;
+ std::cout << "Starting tests... " << ac<< std::endl;
- MWEParser parser;
+ MWEIndex temp_index;
+
+ MWEParser parser(temp_index);
parser.parse_file(av[1]);
}
--
GitLab