Merge pull request #18 from j0tunn/feat/prj.13

ДЗ-13 (12). Недвижимость

common/CMakeLists.txt

@@ -16,7 +16,7 @@ function(setupGTest target)
   FetchContent_MakeAvailable(googletest)
 
   include_directories(${googletest_SOURCE_DIR}/googlemock/include)
-  target_link_libraries(${target} GTest::gtest_main GTest::gmock_main)
+  target_link_libraries(${target} PUBLIC GTest::gtest_main GTest::gmock_main)
 
   include(GoogleTest)
   gtest_discover_tests(${target})

projects/13/CMakeLists.txt

@@ -0,0 +1,36 @@
+cmake_minimum_required(VERSION 3.10)
+
+include(${CMAKE_CURRENT_SOURCE_DIR}/../../common/CMakeLists.txt)
+
+project(fashion_mnist VERSION ${PROJECT_VERSION})
+
+add_library(utils_lib
+    utils.cpp
+)
+
+# catboost
+include_directories(${CMAKE_CURRENT_BINARY_DIR})
+target_link_directories(utils_lib PUBLIC ${CMAKE_CURRENT_BINARY_DIR}/catboost/static)
+target_link_libraries(utils_lib PUBLIC catboostmodel)
+
+add_executable(${PROJECT_NAME}
+    main.cpp
+)
+target_link_libraries(${PROJECT_NAME} utils_lib)
+
+# Tests
+add_executable(test
+    model_test.cc
+    fixtures
+)
+target_link_libraries(test PUBLIC utils_lib)
+
+add_custom_command(
+    OUTPUT fixtures
+    COMMAND ${CMAKE_COMMAND} -E create_symlink ${CMAKE_CURRENT_SOURCE_DIR}/fixtures ${CMAKE_CURRENT_BINARY_DIR}/fixtures
+)
+
+setupGTest(test)
+
+# Package
+setupCPack(${PROJECT_NAME})

projects/13/catboost-fix.diff

@@ -0,0 +1,43 @@
+diff --git a/catboost/libs/model/ya.make b/catboost/libs/model/ya.make
+index 7f97b04542..f5a627da48 100644
+--- a/catboost/libs/model/ya.make
++++ b/catboost/libs/model/ya.make
+@@ -3,8 +3,8 @@ LIBRARY()
+
+
+ SRCS(
+-    GLOBAL cpu/formula_evaluator.cpp
+-    GLOBAL model_import_interface.cpp
++    cpu/formula_evaluator.cpp
++    model_import_interface.cpp
+     cpu/evaluator_impl.cpp
+     cpu/quantization.cpp
+     ctr_data.cpp
+diff --git a/library/cpp/logger/ya.make b/library/cpp/logger/ya.make
+index 14dfaf820a..956532ef13 100644
+--- a/library/cpp/logger/ya.make
++++ b/library/cpp/logger/ya.make
+@@ -17,12 +17,12 @@ SRCS(
+     GLOBAL composite_creator.cpp
+     element.cpp
+     file.cpp
+-    GLOBAL file_creator.cpp
++    file_creator.cpp
+     filter.cpp
+     filter_creator.cpp
+     log.cpp
+     null.cpp
+-    GLOBAL null_creator.cpp
++    null_creator.cpp
+     priority.h
+     record.h
+     rotating_file.cpp
+@@ -35,7 +35,7 @@ SRCS(
+     GLOBAL system_creator.cpp
+     thread.cpp
+     thread_creator.cpp
+-    GLOBAL uninitialized_creator.cpp
++    uninitialized_creator.cpp
+     reopen.h
+ )
+

projects/13/main.cpp

@@ -0,0 +1,36 @@
+#include <iostream>
+#include <string>
+#include <fstream>
+#include <catboost/wrapped_calcer.h>
+#include "utils.h"
+
+using namespace std;
+
+int main(int argc, char* argv[]) {
+    if (argc != 3) {
+        cerr << "Usage: fashion_mnist <test-file> <model-file>\n";
+        return 1;
+    }
+
+    const string testFile(argv[1]);
+    const string modelFile(argv[2]);
+
+    ModelCalcerWrapper calcer(modelFile);
+    ifstream testData(testFile);
+
+    unsigned total = 0;
+    unsigned correct = 0;
+    for (string line; getline(testData, line);) {
+        const auto [expected, features] = parseTestData(line);
+        const auto actual = predict(calcer, features);
+
+        ++total;
+        if (actual == expected) {
+            ++correct;
+        }
+    }
+
+    cout << static_cast<double>(correct)/total << endl;
+
+    return 0;
+}

projects/13/model_test.cc

@@ -0,0 +1,18 @@
+#include <fstream>
+#include <iostream>
+#include <gtest/gtest.h>
+#include <catboost/wrapped_calcer.h>
+
+#include "utils.h"
+
+using namespace std;
+
+TEST(CatboostClassifier, testModel) {
+    ModelCalcerWrapper calcer("fixtures/model.cbm");
+    ifstream testData{"fixtures/test_data_catboost.txt"};
+
+    for (string line; getline(testData, line);) {
+        auto [expected, features] = parseTestData(line);
+        EXPECT_EQ(predict(calcer, features), expected);
+    }
+}

projects/13/prebuild

@@ -0,0 +1,18 @@
+#!/bin/bash
+
+mkdir deps
+cd deps
+
+git clone --depth 1 --branch v1.1.1 https://github.com/catboost/catboost.git
+cd catboost
+
+# fix static build
+# https://github.com/catboost/catboost/issues/2187
+git apply ../../catboost-fix.diff
+
+# Looks like by default c-object build with -fPIC and cxx-objects with -fPIE
+# build with the same flags
+./ya make -DCXXFLAGS=-fPIE -DCFLAGS=-fPIE -r catboost/libs/model_interface/static
+
+cd ../../
+ln -s deps/catboost/catboost/libs/model_interface catboost

projects/13/utils.cpp

@@ -0,0 +1,41 @@
+#include <sstream>
+#include <algorithm>
+#include <iterator>
+#include <locale>
+#include <memory>
+#include "utils.h"
+
+using namespace std;
+
+struct CommaIsSpace : ctype<char> {
+    CommaIsSpace() : ctype<char>(get_table()) {}
+
+    static mask const* get_table() {
+        static mask rc[table_size];
+
+        rc[','] = ctype_base::space;
+        rc[' '] = ctype_base::space;
+        rc['\n'] = ctype_base::space;
+
+        return &rc[0];
+    }
+};
+
+pair<unsigned, vector<float> > parseTestData(const string& src) {
+    istringstream lineStream(src);
+    lineStream.imbue(locale(lineStream.getloc(), new CommaIsSpace));
+
+    unsigned expected;
+    lineStream >> expected;
+
+    vector<float> features;
+    copy(istream_iterator<float>(lineStream), istream_iterator<float>(), back_inserter(features));
+
+    return make_pair(move(expected), move(features));
+}
+
+unsigned predict(ModelCalcerWrapper& calcer, const vector<float>& features) {
+    auto res = calcer.CalcMulti(features);
+    return distance(res.begin(), max_element(res.begin(), res.end()));
+}
+

projects/13/utils.h

@@ -0,0 +1,9 @@
+#pragma once
+
+#include <vector>
+#include <string>
+#include <utility>
+#include <catboost/wrapped_calcer.h>
+
+std::pair<unsigned, std::vector<float> > parseTestData(const std::string& inputStream);
+unsigned predict(ModelCalcerWrapper& calcer, const std::vector<float>& features);

projects/current

@@ -1 +1 @@
-12
+13