安装

安装 adbase 需要的系统环境：

gcc 4.8+
cmake 2.6.4+
libevent-2.0.22+
librdkafka-0.9.0+ (非必需)
Note
librdkafka 只有编译生成 libadbase_kafka 时需要该依赖，如果不使用 AIMS 相关模块，可以不依赖该库

安装 gcc 4.8+

如果系统中有 gcc 4.8+ 的编译环境可以跳过本节

安装过程

下载 gcc 4.8 的源码
wget http: //ftp.gnu.org/gnu/gcc/gcc-4.8.4/gcc-4.8.4.tar.bz2
tar -jxvf gcc-4.8.4.tar.bz2

下载编译所需依赖库
cd gcc-4.8.0
./contrib/download_prerequisites
cd ..

进入此目录，执行以下命令，生成 makefile 文件
cd build
../configure --prefix=/usr/local/adinf/gcc4.8 --enable-checking=release --enable-languages=c,c++
# 如果不要32bit交叉编译环境可以加--disable-multilib 编译选项

编译
# j 后面的是核心数，编译速度会比较快
make -j 12

安装
sudo make install

安装编译常见问题

报错gnu/stubs-32.h no such file or directory
解决办法：命令 yum install glibc-devel.i686问题解决

安装 adbase

下载 adbase 源码
git clone git@github.com:weiboad/adbase.git

配置项目
[zhongxiu@bpdev adbase_v2]$ ./cmake.sh
Start cmake configure.....
编译 Debug 级别 [Debug(D)|Release(R)]: R # D -> debug R -> release
编译环境[32bit(32)|64bit(64)]: 64
安装 adbase_kafka 模块 [Y|N]: y
是否是开发模式 [Y|N]: n

Note
如果使用 AIMS 相关模块请安装 adbase_kafka
编译并安装
cd build
make -j 12
make install

生成第一个 C++ 项目

adbase 提供 seed 项目模块可以自动生成项目的基础代码，达到快速开发，开发人员仅需配置需要的项目可选模块即可生成, 在本次 demo 中准备编写一个字符串大小写转化的 Server, 并且创建一个每秒打印 "hello word" 的定时器, 通过分解项目得出需要两个模块分别是 adserver、timer, 暂定项目名称为 hello

创建项目

创建项目目录
mkdir hello
cd hello

创建配置文件 adbase.ini
[project]
; 项目名称
ADINF_PROJECT_NAME=hello
; 项目描述
ADINF_PROJECT_SUMMARY=Weibo adinf hello
; 项目主页
ADINF_PROJECT_URL=http://adinf.weiboad.org
; 项目打包相关信息
ADINF_PROJECT_VENDOR=zhongxiu <zhongxiu@staff.weibo.com>
ADINF_PROJECT_PACKAGER=zhongxiu <zhongxiu@staff.weibo.com>
[module]
; 是否生成 adserver 模块代码
adserver=yes
; 是否生成 timer 模块代码
timer=yes
; 是否生成 aims consumer 模块代码
kafkac=no
; 是否生成 aims producer 模块代码
kafkap=no
; 是否生成 aims logging 模块代码 (强烈推荐开启)
logging=yes
[params]
; 对于 timer 模块该参数生效，定时器名称，多个定时器用逗号分隔
timers=print
; 对于 adserver 模块该参数生效，http server 的 controller 名称, 多个用逗号分隔
http_controllers=Index
; 对于 aims consumer 模块该参数生效，分别是kafka consumer 名称、topic、groupid, 多个用逗号分隔, 三个参数的个数必须一一对应
kafka_consumers=Out
kafka_consumers_topics=test
kafka_consumers_groups=test_cpp
; 对于 aims producer 模块该参数生效，分别是kafka producer 名称、topic, 多个用逗号分隔, 两个参数的个数必须一一对应
kafka_producers=In
kafka_producers_topics=test
[files]
src/main.cpp=src/@ADINF_PROJECT_NAME@.cpp
rpm/main.spec.in=rpm/@ADINF_PROJECT_NAME@.spec.in
[execs]
cmake.sh=1
build_rpm.in=1

生成项目
[zhongxiu@bpdev hello]$ /usr/local/bin/adbase_skeleton
20160718 10:33:24.402927Z 139970968688992 INFO ./ - Seed.cpp:714
Generate file list:
./rpm/hello.spec.in
./rpm/build_rpm.in
./src/Timer.hpp
./src/BootStrap.cpp
./src/HeadProcessor.cpp
./src/AdbaseConfig.hpp
./src/McProcessor.hpp
./src/hello.cpp
./src/Http.hpp
./src/AdServer.hpp
./src/McProcessor.cpp
./src/HeadProcessor.hpp
./src/AdServer.cpp
./src/Http/Index.hpp
./src/Http/Server.cpp
./src/Http/HttpInterface.hpp
./src/Http/Server.hpp
./src/Http/Index.cpp
./src/Http/HttpInterface.cpp
./src/Timer.cpp
./src/BootStrap.hpp
./src/CMakeLists.txt
./src/App.hpp
./src/App.cpp
./src/Http.cpp
./src/Version.hpp.in
./conf/system.ini
./cmake.sh
./CMakeLists.txt

实现定时器输出 hello world

adbase 在生成定时器代码默认会生成 Timer.cpp Timer.hpp 两个文件，Timer 类中的 print 是本次项目在生成前配置的该项目的一个定时器，所以仅需在该回调函数中实现定时器逻辑即可

修改 Timer.cpp
void Timer::print(void*) {
LOG_INFO << "Hello world";
}

修改配置文件将时间间隔设置为 1000ms
[timer]
; 单位ms
intervalPrint = 1000
[logging]
logsDir=logs/
; 日志分卷大小
logRollSize=52428800
; 1: LOG_TRACE 2: LOG_DEBUG 3: LOG_INFO
logLevel=1
; 将日志异步写关闭，以便调试
isAsync=no

Note
将日志异步写关闭，以便调试 isAsync=no
编译运行
[zhongxiu@bpdev hello]$ ./cmake.sh
Start cmake configure.....
编译 Debug 级别 [Debug(D)|Release(R)]: D
....
[zhongxiu@bpdev build]$ make -j 12
Scanning dependencies of target hello
[ 9%] [ 18%] [ 36%] [ 36%] [ 45%] [ 54%] [ 63%] [ 72%] Building CXX object bin/CMakeFiles/hello.dir/BootStrap.o
Building CXX object bin/CMakeFiles/hello.dir/hello.o
Building CXX object bin/CMakeFiles/hello.dir/App.o
Building CXX object bin/CMakeFiles/hello.dir/Http.o
Building CXX object bin/CMakeFiles/hello.dir/Http/Server.o
Building CXX object bin/CMakeFiles/hello.dir/Http/HttpInterface.o
Building CXX object bin/CMakeFiles/hello.dir/AdServer.o
[ 81%] [ 90%] Building CXX object bin/CMakeFiles/hello.dir/Http/Index.o
[100%] Building CXX object bin/CMakeFiles/hello.dir/McProcessor.o
Building CXX object bin/CMakeFiles/hello.dir/Timer.o
Building CXX object bin/CMakeFiles/hello.dir/HeadProcessor.o
[100%] Built target hello
[zhongxiu@bpdev build]$ ./bin/hello -c ../conf/system.ini
20160718 18:46:25.122239 140572416194912 TRACE initHttp Init http server, host:0.0.0.0 port:10010 - AdServer.cpp:87
20160718 18:46:25.122375 140572416194912 DEBUG Server Init Http Server. - Server.cpp:46
20160718 18:46:25.122412 140572416194912 DEBUG registerLocation Server Location: /server/status register success. - Server.cpp:234
20160718 18:46:25.122430 140572416194912 DEBUG registerLocation Server Location: /index/index register success. - Server.cpp:234
20160718 18:46:25.122483 140572416194912 DEBUG Acceptor Bind: 0.0.0.0:10011 - Acceptor.cpp:29
20160718 18:46:25.122669 140572405700352 DEBUG threadFunc Start create base event. - Server.cpp:116
20160718 18:46:25.122764 140572405700352 TRACE threadFunc Worker start. - Server.cpp:130
20160718 18:46:25.122783 140572405700352 DEBUG threadFunc Start create new evhttp. - Server.cpp:132
....
20160718 18:46:28.154159 140572416194912 INFO Hello world - Timer.cpp:58
20160718 18:46:29.154325 140572416194912 INFO Hello world - Timer.cpp:58
20160718 18:46:30.154650 140572416194912 INFO Hello world - Timer.cpp:58
20160718 18:46:31.155244 140572416194912 INFO Hello world - Timer.cpp:58
20160718 18:46:32.155849 140572416194912 INFO Hello world - Timer.cpp:58
20160718 18:46:33.156313 140572416194912 INFO Hello world - Timer.cpp:58
20160718 18:46:34.156995 140572416194912 INFO Hello world - Timer.cpp:58
20160718 18:46:35.157595 140572416194912 INFO Hello world - Timer.cpp:58
20160718 18:46:36.158060 140572416194912 INFO Hello world - Timer.cpp:58
通过查看日志完成了每个一秒打印一行 hello world 的功能

实现大小写转化 Server

默认 adserver 会同时支持 http / memecache / adhead ，在这个例子中将全部实现

实现 Http server
实现http server 大小写转化接口在 Index 的 controller 中实现 convert Action 函数, 请求访问地址 /index/convert, 请求参数 method=lower 转化为小写, method=upper 转化为大写，data=数据. 例如 /index/convert?method=upper&data=abc
```
- 修改 Http/Index.hpp
```
void convert(adbase::http::Request* request, adbase::http::Response* response, void*);

修改 Http/Index.cpp
// {{{ void Index::convert()
void Index::convert(adbase::http::Request* request, adbase::http::Response* response, void*) {
std::string method = request->getQuery("method");
bool isLower = false;
if (method.compare("lower") == 0) {
isLower = true;
}
std::string data = request->getQuery("data");
if (isLower) {
std::transform(data.begin(), data.end(), data.begin(), ::tolower);
} else {
std::transform(data.begin(), data.end(), data.begin(), ::toupper);
}
responseJson(response, data, 0, "", true);
}
// }}}
修改 Http/Index.cpp
// {{{ void Index::registerLocation()
void Index::registerLocation(adbase::http::Server* http) {
ADSERVER_HTTP_ADD_API(http, Index, index);
ADSERVER_HTTP_ADD_API(http, Index, convert);
}
// }}}
编译运行并测试
[zhongxiu@bpdev adbase_v2]$ curl '127.0.0.1:10010/index/convert?method=lower&data=ABC'
{"code":0,"baseid":290655402196992,"data":"abc","msg":""}
[zhongxiu@bpdev adbase_v2]$

实现 Memcache server
实现Memcache 协议的计算，在当前大小写转化的需求中可以使用 memecache 的 get 命令，加入定位用 # 号分隔操作参数，如 lower::ABC 代表将 ABC 字符串转化为小写
- 修改 McProcessor.cpp
  // {{{ adbase::mc::ProtocolBinaryResponseStatus McProcessor::get()
  adbase::mc::ProtocolBinaryResponseStatus McProcessor::get(const void* key,
  uint16_t keylen,
  adbase::Buffer *data) {
  std::string keyData(static_cast<const char*>(key), static_cast<size_t>(keylen));
  std::vector<std::string> infos = adbase::explode(keyData, '#', true);
  if (infos.size() != 2) {
  return adbase::mc::PROTOCOL_BINARY_RESPONSE_NOT_SUPPORTED;
  }
  bool isLower = false;
  if (infos[0].compare("lower") == 0) {
  isLower = true;
  }
  std::string convertStr = infos[1];
  if (isLower) {
  std::transform(convertStr.begin(), convertStr.end(), convertStr.begin(), ::tolower);
  } else {
  std::transform(convertStr.begin(), convertStr.end(), convertStr.begin(), ::toupper);
  }
  data->append(static_cast<const char*>(convertStr.c_str()), static_cast<size_t>(convertStr.size()));
  LOG_INFO << "Mc GET key:" << keyData;
  return adbase::mc::PROTOCOL_BINARY_RESPONSE_SUCCESS;
  }
  // }}}
- 编译运行并测试
  [zhongxiu@bpdev adbase_v2]$ telnet 127.0.0.1 10011
  Trying 127.0.0.1...
  Connected to 127.0.0.1.
  Escape character is '^]'.
  get lower#ABC
  VALUE lower#ABC 0 3
  abc
  END
  get upper#eeee
  VALUE upper#eeee 0 4
  EEEE
  END
  get errorkey
  VALUE errorkey 0 0
  END

实现 Adhead server
Adhead 基本和 Memcache 一致通过发送 lower::ABC 代表将 ABC 转化为小写
- 修改 HeadProcessor.cpp
  // {{{ adbase::head::ProtocolBinaryResponseStatus HeadProcessor::readHandler()
  adbase::head::ProtocolBinaryResponseStatus HeadProcessor::readHandler(adbase::head::ProtocolBinaryDataType datatype,
  const void *data,
  ssize_t datalen,
  adbase::head::ProtocolBinaryDataType* resDataType,
  adbase::Buffer* responseData) {
  (void)datatype;
  std::string msgData(static_cast<const char*>(data), static_cast<size_t>(datalen));
  std::vector<std::string> infos = adbase::explode(msgData, '#', true);
  if (infos.size() != 2) {
  return adbase::head::PROTOCOL_BINARY_RESPONSE_NOT_SUPPORTED;
  }
  bool isLower = false;
  if (infos[0].compare("lower") == 0) {
  isLower = true;
  }
  std::string convertStr = infos[1];
  if (isLower) {
  std::transform(convertStr.begin(), convertStr.end(), convertStr.begin(), ::tolower);
  } else {
  std::transform(convertStr.begin(), convertStr.end(), convertStr.begin(), ::toupper);
  }
  *resDataType = adbase::head::PROTOCOL_BINARY_TYPE_ASCII;
  responseData->append(convertStr);
  return adbase::head::PROTOCOL_BINARY_RESPONSE_SUCCESS;
  }
  // }}}
- 编译运行并测试
  // 通过 php adhead 测试如下：
  use \Adinf\Adhead\Factory as AdheadFactory;
  use \Adinf\Adhead\Protocol\Protocol as AdheadProtocol;
  $start = microtime(true);
  AdheadFactory::init();
  $adhead = AdheadFactory::adhead(AdheadFactory::ADHEAD_SYNC, "127.0.0.1", 10012);
  $adhead->setRecvTimeoutSec(100);
  $input = 'lower#ABC';
  $adhead->write(11, $input, AdheadProtocol::DATATYPE_JSON);
  $data = $adhead->read();
  var_dump($data);
  $input = 'upper#eeee';
  $adhead->write(11, $input, AdheadProtocol::DATATYPE_JSON);
  $data = $adhead->read();
  var_dump($data);
  $end = microtime(true);
  var_dump($end - $start);
  Result:
  array(3) {
  'type' =>
  int(0)
  'status' =>
  int(0)
  'data' =>
  string(3) "abc"
  }
  array(3) {
  'type' =>
  int(0)
  'status' =>
  int(0)
  'data' =>
  string(4) "EEEE"
  }
  double(0.0049090385437012)

Table of Contents