PoliqarpReader

CMakeLists.txt

@@ -68,7 +68,6 @@ add_subdirectory(libpwrutils)
 add_subdirectory(libcorpus2)
 add_subdirectory(corpus2tools)
 add_subdirectory(tests)
-
 if(CORPUS2_BUILD_POLIQARP)
 	add_subdirectory(poliqarp-library)
 	add_subdirectory(poliqarp)

libcorpus2/io/fastxces.cpp

@@ -5,8 +5,8 @@
 
 namespace Corpus2 {
 
-bool FastXcesReader::registered = TokenReader::register_reader<FastXcesReader>("xces-fast",
-	"ign,loose,strict");
+bool FastXcesReader::registered = TokenReader::register_reader<FastXcesReader>(
+	"xces-fast","ign,loose,strict");
 
 class FastXcesReaderImpl
 {

poliqarp-library/CMakeLists.txt

 project (PoliqarpLibrary)
 cmake_minimum_required(VERSION 2.8.0)
 set(CMAKE_MODULE_PATH ${CMAKE_CURRENT_SOURCE_DIR}/CMakeScripts)
+set(pqlib_ver_major 1)
+set(pqlib_ver_minor 0)
 
 INCLUDE (CheckIncludeFiles)
 INCLUDE (CheckLibraryExists)
@@ -136,14 +138,18 @@ set(poliqarpd_SRC
 	poliqarpd/utils.c
 )
 
-add_library(libpoliqarp SHARED ${foostring_SRC} ${progress_SRC} ${unibits_SRC} ${common_SRC} ${sakura_SRC} ${BF_SOURCES} ${PoliqarpLibrary_BINARY_DIR}/sakura/parser.h)
-add_dependencies(libpoliqarp ${PoliqarpLibrary_BINARY_DIR}/sakura/parser.ha)
+add_library(poliqarpc2 SHARED ${foostring_SRC} ${progress_SRC} ${unibits_SRC} ${common_SRC} ${sakura_SRC} ${BF_SOURCES} ${PoliqarpLibrary_BINARY_DIR}/sakura/parser.h)
+add_dependencies(poliqarpc2 ${PoliqarpLibrary_BINARY_DIR}/sakura/parser.ha)
+set_target_properties(poliqarpc2 PROPERTIES
+	VERSION "${pqlib_ver_major}.${pqlib_ver_minor}"
+	SOVERSION ${pqlib_ver_major})
+
 add_executable(poliqarpc-shared utils/poliqarpc.c)
-target_link_libraries(poliqarpc-shared libpoliqarp pthread)
+target_link_libraries(poliqarpc-shared poliqarpc2 pthread)
 add_executable(poliqarpd-shared ${poliqarpd_SRC})
-target_link_libraries(poliqarpd-shared libpoliqarp pthread)
+target_link_libraries(poliqarpd-shared poliqarpc2 pthread)
 if(UNIX)
-	install(TARGETS libpoliqarp LIBRARY DESTINATION lib)
+	install(TARGETS poliqarpc2 LIBRARY DESTINATION lib)
 	install(DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/sakura DESTINATION include
 		FILES_MATCHING PATTERN "*.h")
 	install(DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/progress DESTINATION include/sakura

poliqarp/CMakeLists.txt

 PROJECT(Corpus2Poliqarp)
 cmake_minimum_required(VERSION 2.8.0)
+set(c2pq_ver_major 1)
+set(c2pq_ver_minor 0)
 include_directories(${PoliqarpLibrary_SOURCE_DIR}/sakura)
 include_directories(${PoliqarpLibrary_SOURCE_DIR})
 include_directories(${PoliqarpLibrary_BINARY_DIR}/sakura)
 include_directories(${PoliqarpLibrary_BINARY_DIR})
-add_library(corpus2poliqarp SHARED pqclient.cpp)
-
-add_executable(c2pqtest c2pqtest.cpp)
-target_link_libraries(corpus2poliqarp libpoliqarp corpus2)
-target_link_libraries(c2pqtest libpoliqarp corpus2poliqarp corpus2 pwrutils )
-
 
+add_library(corpus2poliqarp SHARED pqclient.cpp pqreader.cpp)
+set_target_properties(corpus2poliqarp PROPERTIES
+	VERSION "${c2pq_ver_major}.${c2pq_ver_minor}"
+	SOVERSION ${c2pq_ver_major})
+target_link_libraries(corpus2poliqarp poliqarpc2 corpus2)
 
+add_executable(c2pqtest c2pqtest.cpp)
+target_link_libraries(c2pqtest poliqarpc2 corpus2poliqarp corpus2 pwrutils )
 
+if(UNIX)
+	install(TARGETS corpus2poliqarp LIBRARY DESTINATION lib)
+	install(TARGETS c2pqtest RUNTIME DESTINATION bin)
+endif(UNIX)

poliqarp/c2pqtest.cpp

@@ -15,7 +15,10 @@ int main(int argc, char** argv)
 	pqc.execute_query();
 	boost::shared_ptr<Corpus2::TokenWriter> writer;
 	writer = Corpus2::TokenWriter::create_stream_writer("plain", std::cout, tagset);
-	while (Corpus2::Token* t = pqc.get_next_focus_token()) {
-		writer->write_token_dispose(t);
+	//while (Corpus2::Token* t = pqc.get_next_focus_token()) {
+	//	writer->write_token_dispose(t);
+	//}
+	while (Corpus2::Sentence::Ptr s = pqc.get_next_match_sequence()) {
+		writer->write_sentence(*s);
 	}
 }

poliqarp/pqclient.cpp

 #include "pqclient.h"
 #include <boost/make_shared.hpp>
+#include <boost/lexical_cast.hpp>
 
 extern "C" {
 	void async_notify_new_results(void* session)
@@ -136,7 +137,7 @@ Token* PoliqarpClient::get_token(size_t pos)
 	poliqarp_get_interpretation_set_info(&set, &sinfo);
 
 	std::auto_ptr<Token> res(new Token());
-	if (!info.space_before) {
+	if (info.space_before) {
 		res->set_wa(PwrNlp::Whitespace::Space);
 	}
 	res->set_orth_utf8(info.text);
@@ -151,25 +152,25 @@ Token* PoliqarpClient::get_token(size_t pos)
 	return res.release();
 }
 
-Sentence::Ptr PoliqarpClient::get_next_sequence(bool whole_sentence)
+boost::shared_ptr<Chunk> PoliqarpClient::get_next_document()
 {
-	Sentence::Ptr sentence;
-	if (info_.used > 0) {
-		if (buffer_pos_ < info_.used) {
-			struct poliqarp_match poli_match;
-			poliqarp_get_match(&buffer_, &poli_match, buffer_pos_++);
-			curr_chunk_doc_id_ = poli_match.document;
-			if (whole_sentence) {
-				//sentence = get_token_range(poli_match.withinStart, poli_match.withinEnd);
+	poliqarp_match match;
+	boost::shared_ptr<Chunk> chunk;
+	if (next_match(match)) {
+		chunk = boost::make_shared<Chunk>();
+		size_t document_id = match.document;
+		chunk->set_attribute("id", "ch" + boost::lexical_cast<std::string>(document_id));
+		chunk->append(get_token_range(match.start, match.end));
+		while (next_match(match)) {
+			if (match.document == document_id) {
+				chunk->append(get_token_range(match.start, match.end));
 			} else {
-				sentence = get_token_range(poli_match.start, poli_match.end);
+				buffer_pos_--;
+				break;
 			}
-		} else {
-			execute_query();
-			sentence = get_next_sequence(whole_sentence);
 		}
 	}
-	return sentence;
+	return chunk;
 }
 
 Sentence::Ptr PoliqarpClient::get_token_range(size_t from, size_t to)

poliqarp/pqclient.h

@@ -32,12 +32,11 @@ public:
 
 	Token* get_next_focus_token();
 	Sentence::Ptr get_next_match_sequence();
+	boost::shared_ptr<Chunk> get_next_document();
 
 	Token* get_token(size_t pos);
 	Sentence::Ptr get_token_range(size_t from, size_t to);
 
-	Sentence::Ptr get_next_sequence(bool whole_sentence);
-
 	size_t get_count_of_matches_so_far();
 	size_t only_count_results();
 	size_t get_corpus_size() const;

poliqarp/pqreader.cpp

+#include "pqreader.h"
+#include "pqclient.h"
+
+namespace Corpus2 {
+
+bool PoliqarpReader::registered = TokenReader::register_path_reader<PoliqarpReader>(
+	"poliqarp","token,chunk,sentence");
+
+
+PoliqarpReader::PoliqarpReader(const Tagset &tagset, const std::string &filename)
+	: TokenReader(tagset), pq_(new PoliqarpClient(tagset, filename)),
+	executed_(false), mode_(PQ_SENTENCES)
+{
+	pq_->compile_query("[]+ within s");
+}
+
+PoliqarpReader::~PoliqarpReader()
+{
+}
+
+void PoliqarpReader::set_query(const std::string &query)
+{
+	pq_->compile_query(query);
+	mode_ = PQ_MANUAL;
+	executed_ = false;
+}
+
+void PoliqarpReader::execute()
+{
+	pq_->execute_query();
+	executed_ = true;
+}
+
+Token* PoliqarpReader::get_next_token()
+{
+	if (!executed_) execute();
+	return pq_->get_next_focus_token();
+}
+
+Sentence::Ptr PoliqarpReader::get_next_sentence()
+{
+	if (!executed_) execute();
+	return pq_->get_next_match_sequence();
+}
+
+boost::shared_ptr<Chunk> PoliqarpReader::get_next_chunk()
+{
+	if (!executed_) execute();
+	return pq_->get_next_document();
+}
+
+void PoliqarpReader::set_option(const std::string &option)
+{
+	if (option == "chunk") {
+		pq_->compile_query("[]+ within p");
+		mode_ = PQ_CHUNKS;
+		executed_ = false;
+	} else if (option == "token") {
+		pq_->compile_query("[]+ within s");
+		mode_ = PQ_SENTENCES;
+		executed_ = false;
+	} else if (option == "token") {
+		pq_->compile_query("[]");
+		mode_ = PQ_TOKENS;
+		executed_ = false;
+	} else {
+		TokenReader::set_option(option);
+	}
+}
+
+std::string PoliqarpReader::get_option(const std::string& option) const
+{
+	return TokenReader::get_option(option);
+}
+
+} /* end ns Corpus2 */

poliqarp/pqreader.h

+#ifndef LIBSORPUS2_PQREADER_H
+#define LIBCORPUS2_PQREADER_H
+
+#include <libcorpus2/io/reader.h>
+#include <deque>
+#include <boost/scoped_ptr.hpp>
+
+
+namespace Corpus2 {
+
+// fwd decl
+class PoliqarpClient;
+
+class PoliqarpReader: public TokenReader
+{
+public:
+	PoliqarpReader(const Tagset& tagset, const std::string& filename);
+
+	enum PQ_MODE {
+		PQ_TOKENS,
+		PQ_SENTENCES,
+		PQ_CHUNKS,
+		PQ_MANUAL
+	};
+
+	~PoliqarpReader();
+
+	void set_query(const std::string& query);
+
+	void execute();
+
+	Token* get_next_token();
+
+	Sentence::Ptr get_next_sentence();
+
+	boost::shared_ptr<Chunk> get_next_chunk();
+
+	void set_option(const std::string& option);
+
+	std::string get_option(const std::string& option) const;
+
+	static bool registered;
+
+protected:
+	void ensure_more();
+
+	boost::scoped_ptr<PoliqarpClient> pq_;
+
+	bool executed_;
+
+	PQ_MODE mode_;
+};
+
+} /* end ns Corpus2 */
+
+#endif // LIBCORPUS2_PQREADER_H

swig/CMakeLists.txt

@@ -2,9 +2,10 @@
 
 PROJECT(corpus2SwigWrap)
 
-set(CORPUS2_LIBS corpus2)
-set(PWRUTILS_LIBS pwrutils)
-set(CORPUS2_PWR_LIBS ${CORPUS2_LIBS} ${PWRUTILS_LIBS})
+set(LIBS "corpus2" "pwrutils")
+if (CORPUS2_BUILD_POLIQARP)
+	set(LIBS ${LIBS} "corpus2poliqarp")
+endif (CORPUS2_BUILD_POLIQARP)
 
 include_directories (${corpus2_SOURCE_DIR})
 include_directories (${pwrutils_SOURCE_DIR})
@@ -48,8 +49,7 @@ SET_SOURCE_FILES_PROPERTIES(corpus2.i PROPERTIES CPLUSPLUS ON)
 SET_SOURCE_FILES_PROPERTIES(corpus2.i PROPERTIES SWIG_FLAGS "-includeall" )
 set(SWIG_MODULE_corpus2_EXTRA_DEPS ${I_FILES})
 SWIG_ADD_MODULE(corpus2 python corpus2.i )
-SWIG_LINK_LIBRARIES(corpus2 ${PYTHON_LIBRARIES} ${PWRUTILS_LIBS})
-SWIG_LINK_LIBRARIES(corpus2 ${PYTHON_LIBRARIES} ${CORPUS2_PWR_LIBS})
+SWIG_LINK_LIBRARIES(corpus2 ${PYTHON_LIBRARIES} ${LIBS})
 # -----------------------------------------------------------------------------
 # -----------------------------------------------------------------------------