TensorRT/samples/common/logger.h,logger.cpp,logging.h源碼研讀

TensorRT/samples/common/logger.h,logger.cpp,logging.h源碼研讀

  • 前言
  • logger.h
  • logger.cpp
  • logging.h
  • extern
  • list initialization
  • using
  • member initializer list
  • move constructor
  • time_t, tm
  • 多重繼承的順序
  • (純)虛擬函數
  • nested class
  • friend class
  • opaque handle
  • enum class
  • assert(0)
  • EXIT\_SUCCESS, EXIT\_FAILURE
  • anonymous namespace(unnamed namespace)
  • inline function
  • 參考連結

前言

本篇介紹TensorRT的logger.hlogger.cpplogging.h這三個檔案。logger.h作為與外部聯繫的接口,logging.h則負則log功能的具體實現。

logger.h

/*
 * Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#ifndef LOGGER_H
#define LOGGER_H

#include "logging.h"

extern Logger gLogger;
extern LogStreamConsumer gLogVerbose;
extern LogStreamConsumer gLogInfo;
extern LogStreamConsumer gLogWarning;
extern LogStreamConsumer gLogError;
extern LogStreamConsumer gLogFatal;

void setReportableSeverity(Logger::Severity severity);

#endif // LOGGER_H

logger.h相當於一個與外部溝通的接口,外部的cpp檔如果想要使用與log有關的功能,只要引用logger.h即可。

log功能的具體實現則由logger.cpplogging.h負責。

logger.cpp

/*
 * Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#include "logger.h"
#include "logging.h"

//from logging.h:
//Logger建構子:Logger(Severity severity = Severity::kWARNING)
Logger gLogger{Logger::Severity::kINFO};
//from logging.h:
//LogStreamConsumer建構子:LogStreamConsumer(LogStreamConsumer&& other)
//LOG_VERBOSE函數,定義於logging.h:LogStreamConsumer LOG_VERBOSE(const Logger& logger)
LogStreamConsumer gLogVerbose{LOG_VERBOSE(gLogger)};
LogStreamConsumer gLogInfo{LOG_INFO(gLogger)};
LogStreamConsumer gLogWarning{LOG_WARN(gLogger)};
LogStreamConsumer gLogError{LOG_ERROR(gLogger)};
LogStreamConsumer gLogFatal{LOG_FATAL(gLogger)};
//此處使用了list initialization

//Logger是nvinfer1::ILogger的子類別,繼承了父類別的Severity
void setReportableSeverity(Logger::Severity severity)
{
    //呼叫Logger::setReportableSeverity
    gLogger.setReportableSeverity(severity);
    //呼叫LogStreamConsumer::setReportableSeverity
    gLogVerbose.setReportableSeverity(severity);
    gLogInfo.setReportableSeverity(severity);
    gLogWarning.setReportableSeverity(severity);
    gLogError.setReportableSeverity(severity);
    gLogFatal.setReportableSeverity(severity);
}

logger.cpp借用了logging.h中定義好的類別:LoggerLogStreamConsumerLogger::Severity及函數LOG_VERBOSELOG_INFOLOG_WARNLOG_ERRORLOG_FATAL等來定義logger.h裡宣告的變數。

logging.h

/*
 * Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#ifndef TENSORRT_LOGGING_H
#define TENSORRT_LOGGING_H

#include "NvInferRuntimeCommon.h"
#include 
#include 
#include 
#include 
#include 
#include 
#include 

/*
enum nvinfer1::ILogger::Severity : int
The severity corresponding to a log message.

Enumerator
kINTERNAL_ERROR 	
An internal error has occurred. Execution is unrecoverable.

kERROR 	
An application error has occurred.

kWARNING 	
An application error has been discovered, but TensorRT has recovered or fallen back to a default.

kINFO 	
Informational messages with instructional information.

kVERBOSE 	
Verbose messages with debugging information.
*/
using Severity = nvinfer1::ILogger::Severity;

/*
typedef basic_stringbuf stringbuf;
Stream buffer to read from and write to string objects.
*/
//本身是一個std::stringbuf,可以做為字串的緩衝區
//另外還帶了std::ostream&型別的成員變數,用來為它自身緩衝區裡的字串做輸出
class LogStreamConsumerBuffer : public std::stringbuf
{
public:
    LogStreamConsumerBuffer(std::ostream& stream, const std::string& prefix, bool shouldLog)
        : mOutput(stream)
        , mPrefix(prefix)
        , mShouldLog(shouldLog)
    {
    }

    //move constructor
    LogStreamConsumerBuffer(LogStreamConsumerBuffer&& other)
        : mOutput(other.mOutput)
    {
    }

    ~LogStreamConsumerBuffer()
    {
        //LogStreamConsumerBuffer繼承自std::stringbuf,
        //而std::stringbuf又繼承自std::streambuf,
        //所以這裡可以直接調用std::streambuf::pbase及std::streambuf::pptr
        // std::streambuf::pbase() gives a pointer to the beginning of the buffered part of the output sequence
        //即回傳一個指向輸出序列開頭的指標
        // std::streambuf::pptr() gives a pointer to the current position of the output sequence
        //即回傳一個指向輸出序列中當前位置的指標
        // if the pointer to the beginning is not equal to the pointer to the current position,
        // call putOutput() to log the output to the stream
        //如果一個當前位置並非輸出序列的開頭,便將輸出序列中的東西給打印出來
        if (pbase() != pptr())
        {
            putOutput();
        }
    }

    // synchronizes the stream buffer and returns 0 on success
    // synchronizing the stream buffer consists of inserting the buffer contents into the stream,
    // resetting the buffer and flushing the stream
    // 虛擬函數,在子類別中可以重新定義(override)它
    // 當我們使用父類別的pointer或reference指向一個子類別的物件時,
    // C++能在runtime知道要調用子類別的版本而非父類別的版本
    //(即runtime polymorphism)
    virtual int sync()
    {
        putOutput();
        return 0;
    }
    
    void putOutput()
	{
	    if (mShouldLog)
	    {
	        // prepend timestamp
	        std::time_t timestamp = std::time(nullptr);
	        tm* tm_local = std::localtime(&timestamp);
	        std::cout << "[";
	        std::cout << std::setw(2) << std::setfill('0') << 1 + tm_local->tm_mon << "/";
	        std::cout << std::setw(2) << std::setfill('0') << tm_local->tm_mday << "/";
	        std::cout << std::setw(4) << std::setfill('0') << 1900 + tm_local->tm_year << "-";
	        std::cout << std::setw(2) << std::setfill('0') << tm_local->tm_hour << ":";
	        std::cout << std::setw(2) << std::setfill('0') << tm_local->tm_min << ":";
	        std::cout << std::setw(2) << std::setfill('0') << tm_local->tm_sec << "] ";
	        //LogStreamConsumerBuffer繼承自std::stringbuf,
	        //所以這裡可以直接調用std::stringbuf::str()
	        // std::stringbuf::str() gets the string contents of the buffer
	        //str()用於獲取buffer裡的字串
	        //"<<": Insert formatted output. This operator (<<) applied to an output stream is known as insertion operator.3
	        // insert the buffer contents pre-appended by the appropriate prefix into the stream
	        //將mPrefix及buffer裡的內容插入到mOutput這個ostream中
	        mOutput << mPrefix << str();
	        // set the buffer to empty
	        //將buffer裡的值設為"",即清空buffer
	        str("");
	        // flush the stream
	        //std::ostream::flush(): Flush output stream buffer
	        mOutput.flush();
	    }
	}

    void setShouldLog(bool shouldLog)
    {
        mShouldLog = shouldLog;
    }

private:
    //參考成員變數
    std::ostream& mOutput;
    std::string mPrefix;
    bool mShouldLog;
};

//!
//! \class LogStreamConsumerBase
//! \brief Convenience object used to initialize LogStreamConsumerBuffer before std::ostream in LogStreamConsumer
//!
class LogStreamConsumerBase
{
public:
    //注意這裡initializer list的寫法:我們可以在initializer list調用成員變數的constructor
    //LogStreamConsumerBuffer建構子:LogStreamConsumerBuffer(std::ostream& stream, const std::string& prefix, bool shouldLog)
    LogStreamConsumerBase(std::ostream& stream, const std::string& prefix, bool shouldLog)
        : mBuffer(stream, prefix, shouldLog)
    {
    }

protected:
    LogStreamConsumerBuffer mBuffer;
};

//!
//! \class LogStreamConsumer
//! \brief Convenience object used to facilitate use of C++ stream syntax when logging messages.
//!  Order of base classes is LogStreamConsumerBase and then std::ostream.
//!  This is because the LogStreamConsumerBase class is used to initialize the LogStreamConsumerBuffer member field
//!  in LogStreamConsumer and then the address of the buffer is passed to std::ostream.
//!  This is necessary to prevent the address of an uninitialized buffer from being passed to std::ostream.
//!  Please do not change the order of the parent classes.
//!
//本身是std::ostream,
//還自帶了LogStreamConsumerBuffer(即std::stringbuf型別的成員變數)
//其建構子接受Severity型別的變數severity做為參數,用以表示其輸出訊息的重要程度,越小越重要
//訊息實際輸出與否是由reportableSeverity控制,當severity小於reportableSeverity時才會輸出
class LogStreamConsumer : protected LogStreamConsumerBase, public std::ostream
{
public:
    //! \brief Creates a LogStreamConsumer which logs messages with level severity.
    //!  Reportable severity determines if the messages are severe enough to be logged.
    //此處initializer list的順序:先LogStreamConsumerBase後std::ostream
    LogStreamConsumer(Severity reportableSeverity, Severity severity)
        //LogStreamConsumerBase(stream, prefix, shouldLog)
        : LogStreamConsumerBase(severityOstream(severity), severityPrefix(severity), severity <= reportableSeverity)
        //mBuffer繼承自LogStreamConsumerBase
        //注意這裡是先呼叫LogStreamConsumerBase建構子,
        //使得mBuffer指向一個有效的位置後,
        //才使用mBuffer來初始化ostream
        , std::ostream(&mBuffer) // links the stream buffer with the stream
        , mShouldLog(severity <= reportableSeverity)
        , mSeverity(severity)
    {
    }

    //move constructor
    LogStreamConsumer(LogStreamConsumer&& other)
        : LogStreamConsumerBase(severityOstream(other.mSeverity), severityPrefix(other.mSeverity), other.mShouldLog)
        , std::ostream(&mBuffer) // links the stream buffer with the stream
        , mShouldLog(other.mShouldLog)
        , mSeverity(other.mSeverity)
    {
    }

    void setReportableSeverity(Severity reportableSeverity)
    {
        //如果reportableSeverity大於等於事先設定好的mSeverity,才會開啟log功能
        mShouldLog = mSeverity <= reportableSeverity;
        mBuffer.setShouldLog(mShouldLog);
    }

private:
    //依據傳入的參數severity決定要寫到標準輸出還是標準錯誤
    static std::ostream& severityOstream(Severity severity)
    {
        //決定寫到標準輸出還是標準錯誤
        return severity >= Severity::kINFO ? std::cout : std::cerr;
    }

    static std::string severityPrefix(Severity severity)
    {
        switch (severity)
        {
        //F for FATAL?
        case Severity::kINTERNAL_ERROR: return "[F] ";
        case Severity::kERROR: return "[E] ";
        case Severity::kWARNING: return "[W] ";
        case Severity::kINFO: return "[I] ";
        case Severity::kVERBOSE: return "[V] ";
        default: assert(0); return "";
        }
    }

    bool mShouldLog;
    Severity mSeverity;
};

//! \class Logger
//!
//! \brief Class which manages logging of TensorRT tools and samples
//!
//! \details This class provides a common interface for TensorRT tools and samples to log information to the console,
//! and supports logging two types of messages:
//!
//! - Debugging messages with an associated severity (info, warning, error, or internal error/fatal)
//! - Test pass/fail messages
//!
//! The advantage of having all samples use this class for logging as opposed to emitting directly to stdout/stderr is
//! that the logic for controlling the verbosity and formatting of sample output is centralized in one location.
//!
//! In the future, this class could be extended to support dumping test results to a file in some standard format
//! (for example, JUnit XML), and providing additional metadata (e.g. timing the duration of a test run).
//!
//! TODO: For backwards compatibility with existing samples, this class inherits directly from the nvinfer1::ILogger
//! interface, which is problematic since there isn't a clean separation between messages coming from the TensorRT
//! library and messages coming from the sample.
//!
//! In the future (once all samples are updated to use Logger::getTRTLogger() to access the ILogger) we can refactor the
//! class to eliminate the inheritance and instead make the nvinfer1::ILogger implementation a member of the Logger
//! object.

class Logger : public nvinfer1::ILogger
{
public:
    Logger(Severity severity = Severity::kWARNING)
        : mReportableSeverity(severity)
    {
    }

    //!
    //! \enum TestResult
    //! \brief Represents the state of a given test
    //!
    enum class TestResult
    {
        kRUNNING, //!< The test is running
        kPASSED,  //!< The test passed
        kFAILED,  //!< The test failed
        //放棄,延遲
        kWAIVED   //!< The test was waived
    };

    //!
    //! \brief Forward-compatible method for retrieving the nvinfer::ILogger associated with this Logger
    //! \return The nvinfer1::ILogger associated with this Logger
    //!
    //! TODO Once all samples are updated to use this method to register the logger with TensorRT,
    //! we can eliminate the inheritance of Logger from ILogger
    //!
    nvinfer1::ILogger& getTRTLogger()
    {
        return *this;
    }

    //!
    //! \brief Implementation of the nvinfer1::ILogger::log() virtual method
    //!
    //! Note samples should not be calling this function directly; it will eventually go away once we eliminate the
    //! inheritance from nvinfer1::ILogger
    //!
    //這裡是override TensorRT/include/NvInferRuntimeCommon.h中宣告了純虛擬函數:
    //virtual void log(Severity severity, const char* msg) TRTNOEXCEPT = 0;
    void log(Severity severity, const char* msg) override
    {
        //LogStreamConsumer是std::ostream的子類別,
        //所以可以使用"<<"運算子
        LogStreamConsumer(mReportableSeverity, severity) << "[TRT] " << std::string(msg) << std::endl;
    }

    //!
    //! \brief Method for controlling the verbosity of logging output
    //!
    //! \param severity The logger will only emit messages that have severity of this level or higher.
    //!
    void setReportableSeverity(Severity severity)
    {
        mReportableSeverity = severity;
    }

    //!
    //! \brief Opaque handle that holds logging information for a particular test
    //!
    //! This object is an opaque handle to information used by the Logger to print test results.
    //! The sample must call Logger::defineTest() in order to obtain a TestAtom that can be used
    //! with Logger::reportTest{Start,End}().
    //!
    //用於記錄測試相關資訊的數據結構
    class TestAtom
    {
    public:
        TestAtom(TestAtom&&) = default;

    private:
        //Logger為TestAtom的friend class
        friend class Logger;

        TestAtom(bool started, const std::string& name, const std::string& cmdline)
            : mStarted(started)
            , mName(name)
            , mCmdline(cmdline)
        {
        }

        bool mStarted;
        std::string mName;
        std::string mCmdline;
    };

    //!
    //! \brief Define a test for logging
    //!
    //! \param[in] name The name of the test.  This should be a string starting with
    //!                  "TensorRT" and containing dot-separated strings containing
    //!                  the characters [A-Za-z0-9_].
    //!                  For example, "TensorRT.sample_googlenet"
    //! \param[in] cmdline The command line used to reproduce the test
    //
    //! \return a TestAtom that can be used in Logger::reportTest{Start,End}().
    //!
    //根據傳入的參數name及cmdline新建一個TestAtom物件後回傳
    static TestAtom defineTest(const std::string& name, const std::string& cmdline)
    {
        return TestAtom(false, name, cmdline);
    }

    //!
    //! \brief A convenience overloaded version of defineTest() that accepts an array of command-line arguments
    //!        as input
    //!
    //! \param[in] name The name of the test
    //! \param[in] argc The number of command-line arguments
    //! \param[in] argv The array of command-line arguments (given as C strings)
    //!
    //! \return a TestAtom that can be used in Logger::reportTest{Start,End}().
    //上面defineTest函數的wrapper:將argc及argv解析成正確的cmdline後呼叫上面defineTest函數
    static TestAtom defineTest(const std::string& name, int argc, char const* const* argv)
    {
        auto cmdline = genCmdlineString(argc, argv);
        return defineTest(name, cmdline);
    }

    //!
    //! \brief Report that a test has started.
    //!
    //! \pre reportTestStart() has not been called yet for the given testAtom
    //!
    //! \param[in] testAtom The handle to the test that has started
    //!
    //輸出測試狀態(為running)並將TestAtom物件的mStarted屬性設為true
    static void reportTestStart(TestAtom& testAtom)
    {
        reportTestResult(testAtom, TestResult::kRUNNING);
        assert(!testAtom.mStarted);
        testAtom.mStarted = true;
    }

    //!
    //! \brief Report that a test has ended.
    //!
    //! \pre reportTestStart() has been called for the given testAtom
    //!
    //! \param[in] testAtom The handle to the test that has ended
    //! \param[in] result The result of the test. Should be one of TestResult::kPASSED,
    //!                   TestResult::kFAILED, TestResult::kWAIVED
    //!
    //確認運行狀態後輸出測試結果
    static void reportTestEnd(const TestAtom& testAtom, TestResult result)
    {
        assert(result != TestResult::kRUNNING);
        //確認testAtom.mStarted是unset的
        assert(testAtom.mStarted);
        reportTestResult(testAtom, result);
    }

    //測試結束的情況之一
    static int reportPass(const TestAtom& testAtom)
    {
        reportTestEnd(testAtom, TestResult::kPASSED);
        return EXIT_SUCCESS;
    }

    //測試結束的情況之二
    static int reportFail(const TestAtom& testAtom)
    {
        reportTestEnd(testAtom, TestResult::kFAILED);
        return EXIT_FAILURE;
    }

    //測試結束的情況之三
    static int reportWaive(const TestAtom& testAtom)
    {
        reportTestEnd(testAtom, TestResult::kWAIVED);
        //kWAIVED狀態也算成功?
        return EXIT_SUCCESS;
    }

    //根據參數pass決定要調用reportPass還是reportFail
    static int reportTest(const TestAtom& testAtom, bool pass)
    {
        return pass ? reportPass(testAtom) : reportFail(testAtom);
    }

    Severity getReportableSeverity() const
    {
        return mReportableSeverity;
    }

private:
    //!
    //! \brief returns an appropriate string for prefixing a log message with the given severity
    //!
    static const char* severityPrefix(Severity severity)
    {
        switch (severity)
        {
        case Severity::kINTERNAL_ERROR: return "[F] ";
        case Severity::kERROR: return "[E] ";
        case Severity::kWARNING: return "[W] ";
        case Severity::kINFO: return "[I] ";
        case Severity::kVERBOSE: return "[V] ";
        default: assert(0); return "";
        }
    }

    //!
    //! \brief returns an appropriate string for prefixing a test result message with the given result
    //!
    //將enum class TestResult類型的result轉換為const char*
    static const char* testResultString(TestResult result)
    {
        switch (result)
        {
        case TestResult::kRUNNING: return "RUNNING";
        case TestResult::kPASSED: return "PASSED";
        case TestResult::kFAILED: return "FAILED";
        case TestResult::kWAIVED: return "WAIVED";
        default: assert(0); return "";
        }
    }

    //!
    //! \brief returns an appropriate output stream (cout or cerr) to use with the given severity
    //!
    //依據傳入的參數severity來決定是回傳std::cout或std::cerr
    static std::ostream& severityOstream(Severity severity)
    {
        return severity >= Severity::kINFO ? std::cout : std::cerr;
    }

    //!
    //! \brief method that implements logging test results
    //!
    //輸出測試結果到severityOstream(Severity::kINFO)這個std::ostream&裡
    static void reportTestResult(const TestAtom& testAtom, TestResult result)
    {
        severityOstream(Severity::kINFO) << "&&&& " << testResultString(result) << " " << testAtom.mName << " # "
                                         << testAtom.mCmdline << std::endl;
    }

    //!
    //! \brief generate a command line string from the given (argc, argv) values
    //!
    //解析argc及argv,設定好ss的buffer裡的內容後,取出其值回傳
    static std::string genCmdlineString(int argc, char const* const* argv)
    {
        //std::stringstream : Stream class to operate on strings.
        //std::ostream::operator<< : Insert formatted output
        //std::stringstream::str : returns a string object with a copy of the current contents of the stream.
        std::stringstream ss;
        for (int i = 0; i < argc; i++)
        {
            if (i > 0)
                ss << " ";
            ss << argv[i];
        }
        return ss.str();
    }

    Severity mReportableSeverity;
};

namespace
{

//!
//! \brief produces a LogStreamConsumer object that can be used to log messages of severity kVERBOSE
//!
//! Example usage:
//!
//!     LOG_VERBOSE(logger) << "hello world" << std::endl;
//!
inline LogStreamConsumer LOG_VERBOSE(const Logger& logger)
{
    return LogStreamConsumer(logger.getReportableSeverity(), Severity::kVERBOSE);
}

//!
//! \brief produces a LogStreamConsumer object that can be used to log messages of severity kINFO
//!
//! Example usage:
//!
//!     LOG_INFO(logger) << "hello world" << std::endl;
//!
inline LogStreamConsumer LOG_INFO(const Logger& logger)
{
    return LogStreamConsumer(logger.getReportableSeverity(), Severity::kINFO);
}

//!
//! \brief produces a LogStreamConsumer object that can be used to log messages of severity kWARNING
//!
//! Example usage:
//!
//!     LOG_WARN(logger) << "hello world" << std::endl;
//!
inline LogStreamConsumer LOG_WARN(const Logger& logger)
{
    return LogStreamConsumer(logger.getReportableSeverity(), Severity::kWARNING);
}

//!
//! \brief produces a LogStreamConsumer object that can be used to log messages of severity kERROR
//!
//! Example usage:
//!
//!     LOG_ERROR(logger) << "hello world" << std::endl;
//!
inline LogStreamConsumer LOG_ERROR(const Logger& logger)
{
    return LogStreamConsumer(logger.getReportableSeverity(), Severity::kERROR);
}

//!
//! \brief produces a LogStreamConsumer object that can be used to log messages of severity kINTERNAL_ERROR
//         ("fatal" severity)
//!
//! Example usage:
//!
//!     LOG_FATAL(logger) << "hello world" << std::endl;
//!
inline LogStreamConsumer LOG_FATAL(const Logger& logger)
{
    return LogStreamConsumer(logger.getReportableSeverity(), Severity::kINTERNAL_ERROR);
}

} // anonymous namespace

#endif // TENSORRT_LOGGING_H

LogStreamConsumerBuffer類別繼承自std::stringbuf,管理著字串緩沖區。

LogStreamConsumerBaseLogStreamConsumerBuffer的wrapper。

LogStreamConsumer:繼承自LogStreamConsumerBasestd::ostream,所以本身兼具字串緩沖區及ostream的角色。

Logger:繼承自nvinfer1::ILogger。透過調用LogStreamConsumer物件來進行輸出。

LOG_VERBOSELOG_INFOLOG_WARNLOG_ERRORLOG_FATAL:這些函數的作用是生成相對應severity的LogStreamConsumer物件,被用於logger.cppgLogVerbosegLogInfogLogWarninggLogErrorgLogFatal等變數的定義中。

extern

logger.h中宣告gLoggergLogVerbose時使用了關鍵字extern,更多介紹可參考C extern關鍵字。

list initialization

logger.cpp中定義gLoggergLogVerbose等變數時使用了所謂的list initialization,更多介紹可參考C++ list initialization。

using

logging.h中將nvinfer1::ILogger::Severity取了一個別名為Severity,更多介紹可參考C++ using。

member initializer list

logging.h中使用initializer list對LogStreamConsumerBuffer的成員變數mOutput進行初始化,同樣地,對於LogStreamConsumerBase的成員變數mBuffer,我們一樣使用initializer list進行初始化,更多介紹可參考C++ member initializer list。

move constructor

logging.h中使用了move constructor:

LogStreamConsumerBuffer(LogStreamConsumerBuffer&& other)
    : mOutput(other.mOutput)
{
}

關於move constructor的介紹,詳見C++ move semantics。

time_t, tm

與時間相關的部份己被筆者改成可執行的代碼,可至cpp-code-snippets/time_t_tm_cout.cpp查看。

多重繼承的順序

logging.hLogStreamConsumer繼承自LogStreamConsumerBasestd::ostream兩個父類別,並且在官方註釋中強調這兩個父類別的順序不可被調換,背後原因詳見C++多重繼承的順序。

(純)虛擬函數

logging.h中在定義Logger的成員函數log時,後面加了一個關鍵字override,這是什麼意思呢?

還有在TensorRT/include/NvInferRuntimeCommon.h中定義nvinfer1::ILogger的成員函數log時,在它前面加了一個virtual,並在後面又加了=0,這又是什麼意思呢?詳見C++ (pure)virutal function & abstract class。

nested class

TestAtom類別被定義在Logger類別內,這是所謂的"nested class",詳見C++ nested class。

friend class

TestAtom類別的定義中,有這麼一句:friend class Logger;。這是將Logger這個類別定義為TestAtom類別的friend class,詳見C++ friend class。

opaque handle

在官方註釋中說明TestAtom這個類別是Opaque handle that holds logging information for a particular test,這是什麼意思呢?詳見C++ (opaque) handle。

enum class

Logger類別中定義了一個enum class TestResult,關於enum class,詳見C++ enum class。

assert(0)

LogStreamConsumer類別的成員函數severityPrefix定義中用到了assert(0),這是什麼意思呢?詳見C++ assert(0)。

EXIT_SUCCESS, EXIT_FAILURE

logging.hLogger類別的成員函數reportPassreportFailreportWaive中,使用了EXIT_SUCCESS, EXIT_FAILURE這兩個常數,關於它們的介紹,詳見C++ EXIT_SUCCESS, EXIT_FAILURE。

anonymous namespace(unnamed namespace)

logging.h中使用一個unnamed namespace把LOG_VERBOSE, LOG_INFO, LOG_WARN, LOG_ERROR, LOG_FATAL等函數包起來,為何要這麼做呢?詳見C++ translation unit & anonymous namespace(unnamed namespace)。

inline function

logging.h裡定義了LOG_VERBOSE, LOG_INFO, LOG_WARN, LOG_ERROR, LOG_FATAL等inline function。此處為何要將它們定義為inline的呢?詳見C++ inline function。

參考連結

nvinfer1::ILogger::Severity

std::stringbuf

std::stringbuf::str

std::ostream::flush

std::ostream::operator<<

std::stringstream

std::ostream::operator<<

std::stringstream::str

C extern關鍵字

C++ list initialization

C++ using

C++ member initializer list

C++ move semantics

cpp-code-snippets/time_t_tm_cout.cpp

C++多重繼承的順序

C++ (pure)virutal function & abstract class

C++ nested class

C++ friend class

C++ (opaque) handle

C++ enum class

C++ assert(0)

C++ EXIT_SUCCESS, EXIT_FAILURE

C++ translation unit & anonymous namespace(unnamed namespace)

C++ inline function

你可能感兴趣的:(C++,TensorRT源碼研讀筆記)