init

CMakeLists.txt

@@ -0,0 +1,163 @@
+# Tutorial: http://www.cmake.org/cmake/help/cmake_tutorial.html
+
+cmake_minimum_required(VERSION 2.6)
+project(concordia CXX)
+
+set (CONCORDIA_VERSION_MAJOR 0)
+set (CONCORDIA_VERSION_MINOR 1)
+
+# ============================== #
+# Production paths
+# ============================== #
+
+set (PROD_RESOURCES_DIRECTORY "${concordia_SOURCE_DIR}/prod/resources")
+
+set (PROD_PUDDLE_TAGSET_PATH "${PROD_RESOURCES_DIRECTORY}/puddle/tagset.txt")
+
+# ============================== #
+# Testing paths
+# ============================== #
+
+set (TEST_RESOURCES_DIRECTORY "${concordia_SOURCE_DIR}/tests/resources")
+set (TEST_PUDDLE_TAGSET_PATH "${TEST_RESOURCES_DIRECTORY}/puddle/basic-tagset.txt")
+
+SET(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_PREFIX}/lib")
+
+set(BASE_TARGETS concordia)
+                
+# ================================================
+# Third-party libraries
+# ================================================
+
+# ----------------------------------------------------
+# Regular expression libraries
+# ----------------------------------------------------
+option(WITH_RE2 "Using RE2 regular expression library" OFF)
+message(STATUS "Using RE2 regular expression library ${WITH_RE2}")
+
+if(WITH_RE2)
+    set(HAVE_RE2 1)
+endif(WITH_RE2)
+
+option(WITH_PCRE "Using PCRE regular expression library" ON)
+message(STATUS "Using PCRE regular expression library ${WITH_PCRE}")
+
+if(WITH_PCRE)
+  set(HAVE_PCRE 1)
+endif(WITH_PCRE)
+
+# ----------------------------------------------------
+# Boost
+# ----------------------------------------------------
+set(Boost_USE_STATIC_LIBS OFF)
+set(Boost_USE_STATIC_RUNTIME OFF)
+find_package(Boost COMPONENTS 
+      serialization unit_test_framework system filesystem program_options iostreams REQUIRED)
+
+# ----------------------------------------------------
+# libconfig
+# ----------------------------------------------------
+find_library(LIBCONFIG_LIB NAMES config++ REQUIRED)
+find_path(LIBCONFIG_INCLUDE libconfig.h++)
+
+if(EXISTS ${LIBCONFIG_LIB} AND EXISTS ${LIBCONFIG_INCLUDE})
+  message(STATUS "Found Libconfig")
+  include_directories(${LIBCONFIG_INCLUDE})
+  link_directories(${LIBCONFIG_LIB})
+endif(EXISTS ${LIBCONFIG_LIB} AND EXISTS ${LIBCONFIG_INCLUDE})
+
+# ----------------------------------------------------
+# Snowball stemmer
+# ----------------------------------------------------
+find_library(LIBSTEMMER_LIB NAMES stemmer REQUIRED)
+find_path(LIBSTEMMER_INCLUDE libstemmer.h)
+
+if(EXISTS ${LIBSTEMMER_LIB} AND EXISTS ${LIBSTEMMER_INCLUDE})
+  message(STATUS "Found libstemmer")
+  include_directories(${LIBSTEMMER_INCLUDE})
+  link_directories(${LIBSTEMMER_LIB})
+endif(EXISTS ${LIBSTEMMER_LIB} AND EXISTS ${LIBSTEMMER_INCLUDE})
+
+# ----------------------------------------------------
+# Logging
+# ----------------------------------------------------
+find_library(LOG4CPP_LIB NAMES log4cpp REQUIRED)
+find_path(LOG4CPP_INCLUDE log4cpp/Appender.hh)
+
+if(EXISTS ${LOG4CPP_LIB} AND EXISTS ${LOG4CPP_INCLUDE})
+  message(STATUS "Found Log4cpp")
+  include_directories(${LOG4CPP_INCLUDE})
+  link_directories(${LOG4CPP_LIB})
+endif(EXISTS ${LOG4CPP_LIB} AND EXISTS ${LOG4CPP_INCLUDE})
+
+# ================================================
+# Concordia: Configuration
+# ================================================
+
+configure_file (
+  "${concordia_SOURCE_DIR}/concordia/common/config.hpp.in"
+  "${concordia_SOURCE_DIR}/concordia/common/config.hpp"
+  )
+
+configure_file (
+  "${concordia_SOURCE_DIR}/prod/resources/concordia-config/concordia.cfg.in"
+  "${concordia_SOURCE_DIR}/prod/resources/concordia-config/concordia.cfg"
+  )
+
+configure_file (
+  "${concordia_SOURCE_DIR}/tests/resources/concordia-config/concordia.cfg.in"
+  "${concordia_SOURCE_DIR}/tests/resources/concordia-config/concordia.cfg"
+  )
+
+
+# ================================================
+# Concordia: sub-projects
+# ================================================
+
+set(ALL_DIRECTORIES concordia concordia-console)
+
+include_directories("${concordia_SOURCE_DIR}")
+
+foreach(dir ${ALL_DIRECTORIES})
+  link_directories("${concordia_BINARY_DIR}/${dir}")
+  add_subdirectory(${dir})
+endforeach(dir)
+
+
+    
+# ================================================
+# Tests
+# ================================================
+
+set(TESTS_TARGETS concordia-tests)
+
+add_subdirectory(tests)
+
+ENABLE_TESTING()
+ADD_TEST(FullTest ${CMAKE_CURRENT_BINARY_DIR}/tests/unit-tests/test_runner)
+
+# Doxygen
+find_package(Doxygen)
+if(DOXYGEN_FOUND)
+    FIND_PROGRAM(DOT_TOOL dot DOC "Dot tool from graphviz")
+    SET(DOXYFILE_LATEX ON)
+    SET(DOXYGEN_HAVE_DOT NO)
+    SET(DOXYGEN_QUIET YES)
+        
+    SET(INPUT_FILES)
+    SET(INPUT_FILES "${INPUT_FILES} ${CMAKE_CURRENT_SOURCE_DIR}")
+
+    if(DOT_TOOL)
+        SET(DOXYGEN_HAVE_DOT YES)
+        message(STATUS "Graphviz found")
+    else(DOT_TOOL)
+        message(STATUS "Consider installing Graphviz for call and caller graphs in the documentation")
+    endif(DOT_TOOL)
+    configure_file(${CMAKE_CURRENT_SOURCE_DIR}/Doxyfile.in ${CMAKE_CURRENT_BINARY_DIR}/Doxyfile @ONLY)
+    add_custom_target(doc ALL ${DOXYGEN_EXECUTABLE} ${CMAKE_CURRENT_BINARY_DIR}/Doxyfile
+    WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}
+    COMMENT "Generating API documentation with Doxygen" VERBATIM)
+    
+endif(DOXYGEN_FOUND)
+
+

INSTALL.txt

@@ -0,0 +1,69 @@
+Concordia Installation & Build Manual
+=================================
+
+This file describes how to compile, build
+and install Concordia library.
+
+Requirements
+============
+
+* cmake
+* Boost library
+* Log4cpp
+* libstemmer (Snowball stemming library)
+* (optional) Doxygen
+
+Boost Ubuntu installation
+=========================
+
+sudo apt-get install libboost-dev libboost-serialization-dev libboost-test-dev libboost-filesystem-dev libboost-system-de libboost-program-options-dev libboost-iostreams-dev
+
+Log4cpp Ubuntu installation
+===========================
+
+sudo apt-get install liblog4cpp5-dev
+
+libconfig Ubuntu installation
+=============================
+
+sudo apt-get install libconfig++-dev
+sudo apt-get install libconfig-dev
+
+libstemmer Ubuntu installation
+==============================
+sudo apt-get install libstemmer-dev
+
+Perl-compatible regular expressions (PCRE) Ubuntu installation
+=======================================================
+
+sudo apt-get install libpcre3-dev
+
+Doxygen Ubuntu installation
+=======================================================
+
+sudo apt-get install doxygen
+
+Installation procedure
+======================
+
+mkdir build
+cd build
+cmake -DCMAKE_BUILD_TYPE=RELEASE ..
+make
+make test
+make install
+
+Documentation
+=============
+
+If Doxygen is available, a successful compilation generates documentation data in three
+formats in the build/doc directory.
+
+The man files in doc/man will be installed during installation. Open doc/html/index.html for
+a HTML version of the same documentation. The latex directory contains uncompiled latex
+files. To generate a single pdf file run
+
+cd doc/latex
+make
+
+This should generate a single file called refman.pdf in the same directory.

README.txt

@@ -0,0 +1 @@
+To generate the full documentation follow the steps in INSTALL

cmake.sh

@@ -0,0 +1,3 @@
+#!/bin/sh
+
+cmake -DCMAKE_BUILD_TYPE=RELEASE ..

concordia-runner.sh

@@ -0,0 +1,6 @@
+#!/bin/sh
+
+echo "Running Concordia"
+
+./build/concordia-console/concordia-console -c prod/resources/concordia-config/concordia.cfg
+

run-checkers.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+
+TARGET_DIR=build
+
+./cpplint.py --filter=-legal,-build/namespaces,-whitespace/labels,-build/include_what_you_use,-runtime/int,-readability/streams,-build/include_order `find concordia concordia-console -type f -regextype posix-extended -regex '.*\.(cpp|hpp|h|c)' ! -regex '.*\./build.*' ! -regex '.*concordia/common/config.hpp' ! -regex '.*/(t|tests)/.*'` 2> cpplint-result.txt
+
+cppcheck -D__cplusplus -D__GNUC__=3 -f  --enable=all echo `find . -type d ! -path './.git*' ! -path "./${TARGET_DIR}"'*' | perl -ne 'chomp; print "-I$_ "'` `find . -type f -regextype posix-extended -regex '.*\.(cpp|hpp|h|c)' ! -regex '.*\./build.*'` 2> cppcheck-result.txt

tests/CMakeLists.txt

@@ -0,0 +1,5 @@
+set(ALL_TESTS_METHODS_DIRS unit-tests common)
+
+foreach(dir ${ALL_TESTS_METHODS_DIRS})
+  add_subdirectory(${dir})
+endforeach(dir)

tests/common/CMakeLists.txt

@@ -0,0 +1,3 @@
+add_library(concordia-tests-common SHARED
+  test_resources_manager.cpp
+)

tests/common/test_resources_manager.cpp

@@ -0,0 +1,22 @@
+#include "test_resources_manager.hpp"
+
+#define PUDDLE_TEST_DIRECTORY "puddle"
+#define CONCORDIA_TAGSET_DIRECTORY "concordia-tagset"
+#define CONCORDIA_CONFIG_DIRECTORY "concordia-config"
+
+string TestResourcesManager::getPuddleFilePath(const string & filename) {
+    string result = string(TEST_RESOURCES_DIRECTORY);
+    return result + "/" + PUDDLE_TEST_DIRECTORY +  "/" + filename;
+}
+
+
+string TestResourcesManager::getTestConcordiaConfigFilePath(const string & filename) {
+    string result = string(TEST_RESOURCES_DIRECTORY);
+    return result + "/" + CONCORDIA_CONFIG_DIRECTORY + "/" + filename;
+}
+
+string TestResourcesManager::getProdConcordiaConfigFilePath(const string & filename) {
+    string result = string(PROD_RESOURCES_DIRECTORY);
+    return result + "/" + CONCORDIA_CONFIG_DIRECTORY + "/" + filename;
+}
+

tests/common/test_resources_manager.hpp

@@ -0,0 +1,21 @@
+#ifndef TEST_RESOURCES_MANAGER_HDR
+#define TEST_RESOURCES_MANAGER_HDR
+
+#include <boost/shared_ptr.hpp>
+#include <string>
+
+#include "concordia/common/config.hpp"
+
+using namespace std;
+
+class TestResourcesManager {
+public:
+    static string getPuddleFilePath(const string & filename);
+
+    static string getTestConcordiaConfigFilePath(const string & filename);
+
+	static string getProdConcordiaConfigFilePath(const string & filename);
+};
+
+#endif
+

tests/resources/concordia-config/concordia-test.cfg

@@ -0,0 +1,9 @@
+#----------------------------
+# Concordia configuration file
+#---------------------------
+#
+
+#Path to the Puddle tagset
+puddle_tagset_path = "puddle/tagset.txt";
+
+### eof

tests/resources/concordia-config/concordia.cfg

@@ -0,0 +1,9 @@
+#----------------------------
+# Concordia configuration file
+#---------------------------
+#
+
+#Path to the Puddle tagset
+puddle_tagset_path = "/home/rafalj/projects/concordia/tests/resources/puddle/basic-tagset.txt";
+
+### eof

tests/resources/concordia-config/concordia.cfg.in

@@ -0,0 +1,9 @@
+#----------------------------
+# Concordia configuration file
+#---------------------------
+#
+
+#Path to the Puddle tagset
+puddle_tagset_path = "@TEST_PUDDLE_TAGSET_PATH@";
+
+### eof

tests/resources/concordia-config/invalid.cfg

@@ -0,0 +1,10 @@
+#----------------------------
+# invalid Concordia configuration file
+#---------------------------
+#
+
+
+#Path to the Puddle tagset
+puddle_tagset_path = "pudlle/tagset.txt" = 1;
+
+### eof

tests/resources/puddle/basic-rules.en.txt

@@ -0,0 +1,44 @@
+# ======================================
+# TIME
+# ======================================
+
+Rule "test 17:00"
+Match: [orth~"([012][0-9]|[1-9]):([0-6][0-9])"];
+Eval:  group(ne_TIME, 1);
+
+# ======================================
+# DATES
+# ======================================
+
+Rule "day number two digits"
+Match: [orth~"(0?[1-9]|[12][0-9]|3[01])"];
+Eval: group(at_DAY_OF_MONTH_NUMBER, 1);
+
+Rule "ORDINAL as day number"
+Match: [orth~"(0?[1-9]|[12][0-9]|3[01]).*" && type~"ORDINAL"];
+Eval: group(at_DAY_OF_MONTH_NUMBER, 1);
+
+Rule "year number four digits"
+Match: [orth~"(19[0-9][0-9]|2[01][0-9][0-9])"];
+Eval: group(at_YEAR_NUMBER, 1);
+
+Rule "date: number MONTH_NAME year"
+Match: [type~"at_DAY_OF_MONTH_NUMBER"] [type~"MONTH_NAME"] [type~"at_YEAR_NUMBER"]?;
+Eval: group(ne_DATE, 1);
+
+# =======================================
+# PERSON
+# =======================================
+
+Rule "person: first_name and upper case"
+Match: [type~"FIRST_NAME"] [orth~"[A-Z].*"];
+Eval: group(ne_PERSON, 1);
+
+# =======================================
+# Testing purposes
+# =======================================
+
+Rule "city: city of XXX"
+Match: [orth~"city"/i] [orth~"of"/i] [type~"CITY"];
+Eval: group(ne_CITY, 3);
+

tests/resources/puddle/basic-tagset.txt

@@ -0,0 +1,7 @@
+[ATTR]
+
+case                    = nom gen dat acc inst loc voc
+
+[POS]
+
+subst = case

tests/resources/puddle/test-rules-bug-1433.en.txt

@@ -0,0 +1,3 @@
+Rule "street"
+Match: [orth~"[P]ort.*"]+;
+Eval: group(ne_ADDRESS, 1);

tests/resources/puddle/test-rules.en.txt

@@ -0,0 +1,15 @@
+Rule "test 17:00"
+Match: [orth~"([012][0-9]|[1-9]):([0-6][0-9])"];
+Eval:  group(ne_TIME, 1);
+
+Rule "test date"
+Match: [orth~"(0?[1-9]|[012][0-9]|3[01])"] [orth~"jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec"/i] [orth~"[12][0-9][0-9][0-9]"];
+Eval:  group(ne_DATE, 1);
+
+Rule "date: MONTH_NAME year"
+Match: [type~"MONTH_NAME"] [orth~"(19[0-9][0-9]|2[01][0-9][0-9])"];
+Eval: group(ne_DATE, 1);
+
+Rule "street"
+Match: [orth~"\d+"] [orth~","]? [orth~"[A-Z].*"]+ [orth~"(St|Rd|Street|Avenue|Lane)"];
+Eval: group(ne_ADDRESS, 1);

tests/unit-tests/CMakeLists.txt

@@ -0,0 +1,7 @@
+
+# executable test_runner
+add_executable(test_runner
+  test_runner.cpp
+  )
+                    
+target_link_libraries(test_runner -Wl,-whole-archive ${TESTS_TARGETS} -Wl,-no-whole-archive ${Boost_LIBRARIES} ${BASE_TARGETS})

tests/unit-tests/test_runner.cpp

@@ -0,0 +1,9 @@
+#define BOOST_TEST_MODULE Main
+#define BOOST_TEST_MAIN
+#include <boost/test/included/unit_test.hpp>
+
+#include "concordia/common/logging.hpp"
+
+BOOST_AUTO_TEST_CASE(TestRunnerMain) {
+    SET_LOGGING_LEVEL("DEBUG");
+}

tests/unit-tests/unit_tests_globals.hpp

@@ -0,0 +1,10 @@
+#ifndef UNIT_TESTS_HDR
+#define UNIT_TESTS_HDR
+
+#define BOOST_TEST_NO_MAIN
+#include <boost/test/unit_test.hpp>
+
+#include "concordia/common/config.hpp"
+#include "concordia/common/logging.hpp"
+
+#endif