Paozhu is a C++ Web development framework with built-in HTTP/1/HTTP/2 ORM, making it easy and elegant to write web programs, breaking the previous inconvenience of writing web programs in C++.
Paozhu has developed the underlying infrastructure for web development, such as multi domain binding, SSL certificate management, cookies, sessions, HTML form management, URL processing, file upload and download, HttpClient, ORM, WebSocket, Logs, HTML Template, covering the basics of web development.
The framework is based on C++20, using new standards with complete compiler support, and modernizing the code. The framework has enabled ASAN and all error prompts(-Wall), making it easy to identify problems during the development phase. It can also be combined with the coredumpctl tool at runtime to view the memory stacks of various programs. Combined with intelligent pointers, the framework has basically eliminated memory leaks and out of bounds issues, and has not encountered any memory leaks or crashes so far.
Paozhu uses C++ mainly to consider the integration of large models (CUDA, file storage, network, memory operations), compared to GO JAVA RUST, which are not very perfect. C++ is also not perfect, RUST syntax is not good-looking, GO is simple but lacks AI culture in Big models, etc. At present, the experience is that ASAN and coredumpctl tools can also ensure the quality of C++, and the new compiler can also provide some hints.
The Paozhu framework uses ASIO's thread pool, which is specifically designed for IO usage, such as network sending and receiving. The framework also has a built-in business thread pool, which is dedicated to running business code.
At present, the frameworks mainly use HTTP and HTTPS, with two threads constantly listening for connections to arrive. When a user accesses using a browser or client, a connection is generated, and after handshake, the server waits for the client to send HTTP header information. Then, the server processes the HTTP protocol. If the URL matches, it registers a function and checks whether the function is a coroutine or a regular function. The coroutine runs directly. If it is a regular function, it is thrown directly into the business thread pool. If the URL is a static file, it is sent directly without going through the business thread.
Because C++coroutines are contagious, when running business code in coroutine registration functions, you need to use ORM coroutine methods. If you cannot initiate coroutine functions in regular functions, you can only use ORM regular functions.
It is generally recommended to use one connection to one thread, and business threads also have queues, so business code should not use the sleep function to sleep casually.
Paozhu supports MacOS Linux and Windows systems.
Uses third-party dependencies
OpenSSL Zlib Brotli ASIO
Optional third-party dependencies
GD qrencode
The framework is based on C++20 or above, ensuring that your system supports compilers that support C++20.
MacOS MacOS 11 BigSur above,suggestion MacOS 14 SonomaUbuntu 22.04 aboveRockyLinux AlmaLinux 9.1 aboveWindows Win10 Compiler 19.25 or above。Download https://github.com/hggq/paozhu
paozhu Config file at conf/server.conf
You can first understand a few variables to get the project running.
[default]
threadmax=1024
threadmin=5
httpport=80
httpsport=443
cothreadnum=8 ;Coroutines run on thread num
http2_enable=1
debug_enable=1
deamon_enable=0
mainhost=www.869869.com
certificate_chain_file=www.869869.com.pem
private_key_file=www.869869.com.key
tmp_dh_file=dh4096.pem
reboot_password=e10adc3949ba59abbe56e057f20f883e ;md5(md5("123456")+"rand_char"+md5("123456"))
reboot_cron =d180h5 ;MDSW+Hhours reboot process M month D day S season (1 4 7 10) W week
clean_cron =m5t600 ;5-minute interval clean 600 seconds ago inactive connection
links_restart_process =n9998877ts1te5 ;More than 15000 connections, restart the process from 1:00 am to 5:00 am
session_type=1 ;session save type 0.file 1.memory 2.redis 3.memcache 4.reserve
static_file_compress_cache=1 ;1 enable, Cache static file compress(gzip,br) content to cache directory
modelspath=/Users/hzq/paozhu/models
serverpath=/Users/hzq/paozhu
viewpath=/Users/hzq/paozhu/view
viewsopath=/Users/hzq/paozhu/module/view
controlpath=/Users/hzq/paozhu/controller
controlsopath=/Users/hzq/paozhu/module/controller
temppath=/Users/hzq/paozhu/temp
logpath=/Users/hzq/paozhu/log
wwwpath=/Users/hzq/paozhu/www/default
pluginspath=/Users/hzq/paozhu/plugins
libspath=/Users/hzq/paozhu/libs
directorylist=1
index=index.html
;usehtmlcache=1
;usehtmlcachetime=3600
staticfile_cache_num=0 ;0 not cache, min 10 begin
rewrite_404=0 ;1 file 2 action url path
rewrite_404_action=index.html
method_pre=
method_after=
show_visitinfo=0
upload_max_size=16777216
siteid=1
groupid=0
alias_domain=
init_func=
[www.869869.com]
wwwpath=/Users/hzq/paozhu/www/default
http2_enable=1
;rewrite_404=1
;rewrite_404_action=index.html|psy/index.html|exam/index.html
;controlsopath=/Users/hzq/paozhu/docs/controller
static_pre=downloadfileauth|upload
method_pre= ;api/dev/hostcors
method_after=
isuse_php=0
rewrite_php=/Users/hzq/www/mpdftest|index.php
fastcgi_host=127.0.0.1 ;/run/php/php8.1-fpm.sock
fastcgi_port=9000
upload_max_size=16777216
siteid=1
groupid=0
alias_domain=
init_func=
threadmax=1024 Maximum number of business threads
threadmin=5 Minimum number of business threads
httpport=80 Server listening port
httpsport=443 SSL server listening port
cothreadnum=8 ;Coroutines run on thread num
http2_enable=1 Whether to enable HTTP/2, each domain can have its own settings
SSL certificate file, after successful application, it should be placed in the conf directory. For production environments, a copy should be placed in /var/local/etc/paozhu.
mainhost=www.869869.com
certificate_chain_file=www.869869.com.pem
private_key_file=www.869869.com.key
reboot_cron Automatic restart process cycle day D week W month M season S h time hour 180 days 5 am restart process
links_restart_process If the number of connections is reached, you can also restart the number of connections after n, The ts start period and te end period will restart if they are idle during this period.
reboot_cron =d180h5 ;MDSW+Hhours reboot process M month D day S season (1 4 7 10) W week
clean_cron =m5t600 ;5-minute interval clean 600 seconds ago inactive connection
links_restart_process =n9998877ts1te5 ;More than 15000 connections, restart the process from 1:00 am to 5:00 am
directorylist=1 Display directory. If the requested static file is a directory name and there is no index.html file in the directory, a directory list will be displayed.
It is recommended to set the production environment to 0
wwwpath=/Users/hzq/paozhu/www/default /Users/hzq/paozhu Suggest changing to your current directory name, Under the project, 'www/default' is a static file directory. If it is a multi domain, it is recommended to create a 'www/domain.com' setting for SAAS services.
upload_max_size=16777216 Upload file size, default is 16M
The following are SAAS settings that can be ignored for now .
siteid=1
groupid=0
alias_domain=
init_func=
themes =
client.get_siteid(); get siteid
client.get_groupid(); get groupid
client.get_theme(); get theme
client.theme_view(contt std::string &); set theme view file
Beginners only need to understand httpport httpsport wwwpath logpath Ensure that the settings are correct, Replace all '/Users/hzq/paozhu' with your directory.
Example of controller/src/techepower.cpp file in project directory
#include "techempower.h"
//@urlpath(null,plaintext)
asio::awaitable<std::string> techempowerplaintext(std::shared_ptr<httppeer> peer)
{
peer->type("text/plain; charset=UTF-8");
peer->set_header("Date", get_gmttime());
peer->output = "Hello, World!";
co_return "";
}
All annotation function cpp files must be placed in thecontroller/srcdirectory.
This techempower.cpp file must contain its own file name #include "techempower.h", The techempower.h file will be automatically generated.
When compiling, use the CMake Hook to first compile an annotation handler that can be used, and then execute this program to handle controller/src Extract functions in the format of //@urlpath(null, plaintext) from all files in the directory, create corresponding header files, and then generate Annotate the function and save the mapping function to the common/autocontrolmethod.hpp file.
Doesn't it have a strong Java Spring Boot flavor so that we can elegantly write URL and function mapping relationships without manually registering functions and writing header files.
//@urlpath(null,plaintext) annotate functions，Mapping with URL，Example http://localhost/plaintext Will access the annotation function of 'plaintext'.
null represents the annotation function that must be executed before accessing plaintext, for example: a backend with permissions that requires verification of login status
//@urlpath(islogin,plaintext) This requires first executing the 'islogin' annotation and returning 'OK' to indicate success.
//@urlpath(null, islogin)
asio::awaitable<std::string> checklogin(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
if(peer->session["userid"].to_int()>0)
{
co_return "ok";
}
client << "You must login";
co_return "";
}
asio::awaitable<std::string> The coroutine annotation function indicates that the code inside does not go through business threads, If heavy calculations are required, it is best to use regular functions so that business threads can handle them without blocking IO threads.
peer->session["userid"] get session content, The content saved after successful login before。
client << "You must login"; It is overloaded with the<<operator, and the actual content is saved in peer->output.
co_return ""; It is a coroutine that returns the result. Please note that it is different from a regular function, and it is best to return an empty string if you are unsure.
The following is a regular registration function, pay attention to the return value, indicating that it goes through the business thread pool, which assigns one thread to each connection. This is a classic application.
It is recommended to use this ordinary function for load computing, otherwise a large amount of computation may affect concurrency, which is possible if it is only for CRUD business
It shouldn't be a problem to use that method. When using coroutines, don't use std::this_thread::sleep_for(std::chrono::microseconds(1000)); This type of code.
//@urlpath(null,hello)
std::string testhello(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
client << " Hello world! 🧨 Paozhu c++ web framework ";
return "";
}
For more code usage cases, please refer to the examples in the controller/src directory
Operate in the root directory of your project.
mkdir build
cd build
cmake ..
make
The compilation is complete, and the server program can now be started. Please pay attention to the execution permissions of the program.
cd ..
MacOS #./bin/paozhu
Linux #sudo ./bin/paozhu
http::obj_val hval;
hval["aaa"]=3344;
hval["bbb"]="1234567890";
std::cout<<"ll:"<<hval["aaa"].to_int()<<std::endl;
std::cout<<"vv:|"<<static_cast<int>(hval["bbb"].get_type())<<"|"<<std::endl;
if(hval["bbb"].is_string())
{
std::cout<<"str:"<<hval["bbb"].to_string()<<std::endl;
std::cout<<"str:"<<hval["bbb"].str_view()<<std::endl;
std::cout<<"str:"<<hval["bbb"].str_view(2,5)<<std::endl;
}
http::obj_val nval;
nval.from_json("{\"bba\":[[[111,222],[333,444],[555,666]],[[777,888],[999,1111],[2222,3333]],[[4444,5555],[6666,7777],[8888,9999]]]}");
std::cout<<"json out:"<<nval.to_json()<<std::endl;
std::string bbb=nval.to_json();
http::obj_val pval;
pval.from_json(bbb);
http::obj_val is a built-in micro object in the framework that can be used to import and export from_json and to_json object conversions.
Does it have a scripting language flavor? It can save values and strings, and can also save tree structures.
http::obj_val is 16bytes size，one KV Object 64bytes size。
hval["aaa"] is KV Object。
http::obj_val Built in methods for objects
set_array() set is array and push(v)set_obj() set is KV and use push(k,v) or obj[key]=valueset_object() set is KV and use push(k,v) or obj[key]=valueis_array() it is an array or notis_obj() it is an object or notis_object() it is an object or notis_string() it is an string or notis_number() it is an number or not client.val["list"].set_array();
obj_val temp;
for (unsigned int i = 0; i < topicm.record.size(); i++)
{
temp["id"] = topicm.record[i].topicid;
temp["parentid"] = topicm.record[i].parentid;
temp["value"] = topicm.record[i].title;
client.val["list"].push(temp);
}
<%c for(auto &a:obj["infos"].as_array()){ %>
<tr>
<td><%c echo<<a["userid"].to_string(); %></td>
<td><%c echo<<a["nickname"].to_string(); %></td>
<td><%c echo<<a["name"].to_string(); %></td>
<td><%c if(a["isopen"].to_int()==1){ %>Open<%c }else{ %>Close<%c } %></td>
<td><a href="/superadmin/edituser?userid=<%c echo<<a["userid"].to_string(); %>">Edit</a> | <a href="/superadmin/deleteuser?userid=<%c echo<<a["userid"].to_string(); %>" onclick="return confirm('Are you sure delete？');">Delete</a></td>
</tr>
<%c } %>
obj_val multi_sort(std::string_view key,unsigned char order); 2D array key name sorting, order is SORT_ASC or SORT_DESC.obj_val multi_sort(std::string_view key,unsigned char order,std::string_view key2,unsigned char order2); 2D array key name sorting, order is SORT_ASC or SORT_DESC, If there is the same secondary sorting,need key2 and order2.
http::obj_val zval4;
zval4.set_array();
float s[6]={83.3,77.75,60.45,83.3,77.75,45.6};
for(unsigned int jjj=0;jjj<6;jjj++)
{
http::obj_val z4;
z4.set_object();
z4["name"]="sss"+std::to_string(jjj);
z4["score"]=s[jjj];
z4["id"]=jjj+10;
zval4.push(z4);
}
http::obj_val zval5=zval4.multi_sort("score",SORT_ASC);
std::cout<<"------"<<std::endl;
if(zval5.is_array())
{
std::cout<<"multi_sort score result: "<<zval5.size()<<std::endl;
for(auto &a:zval5.as_array())
{
if(a.is_object())
{
for(auto &[aa,bb]:a.as_object())
{
std::cout<<"zval5["<<aa<<"]="<< bb.to_string()<<std::endl;
}
}
std::cout<<std::endl;
}
}
http::obj_val zval6=zval4.multi_sort("score",SORT_ASC,"id",SORT_DESC);
std::cout<<"------"<<std::endl;
if(zval6.is_array())
{
std::cout<<"multi_sort score id result: "<<zval6.size()<<std::endl;
for(auto &a:zval6.as_array())
{
if(a.is_object())
{
for(auto &[aa,bb]:a.as_object())
{
std::cout<<"zval6["<<aa<<"]="<< bb.to_string()<<std::endl;
}
}
std::cout<<std::endl;
}
}
std::cout<<"------"<<std::endl;
Mainly used for complex object swapping programs, if it is not a complex object, C++ built-in POD, class objects, or STL library methods can be used.
The difference between Session and Cookie is that session data is stored on the server side, cookie data is stored on the browser side, and cookie data has a limited size of 4Kb.
Example file controller/src/testsessionid.cpp
//@urlpath(null,testsetsession)
std::string testsetsession(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
client << "hello world! add cookie ";
client << client.get_hosturl();
int semid = client.get["aaa"].to_int();
if (semid != 0)
{
client.session["aaa"] = semid;
client << "<br />set session 2222 ";
}
else
{
client.session["aaa"] = 1111;
}
client.save_session();
client << "<p><a href=\"" << client.get_hosturl() << "/testsetsession?aaa=2222\">2222 session</a></p>";
client.set_cookie("ccbb", "123456abc", 7200 * 24);
client.set_cookie("same", "samesite", 7200 * 24, "/", client.host, false, true, "Lax");
client.set_header("server", "cpphttp");
client << "<p><a href=\"" << client.get_hosturl() << "/testshowsession\">show</a></p>";
client << "<p><a href=\"" << client.get_hosturl() << "/testshowsession?sessionid=" << client.get_session_id()
<< "\">session(" << client.get_session_id() << ")</a></p>";
client << "<p>session aaa:|" << client.session.to_json() << "|</p>";
return "";
}
//@urlpath(null,testshowsession)
std::string testshowsession(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
client << "hello world! show cookie<br/>";
client << "<p><a href=\"" << client.get_hosturl() << "/testsetsession\">add cookie</a></p>";
client << "<p>session aaa:|" << client.session.to_json() << "|</p>";
client << "<p>session aaa:|" << client.session["aaa"].to_string() << "|</p>";
client << "<p>session_id:|" << client.get_session_id() << "|</p>";
client << "<p>cookie same:" << client.cookie["same"] << "</p>";
client << "<p>cookie ccbb:" << client.cookie["ccbb"] << "</p>";
std::string semid = client.get["sessionid"].as_string();
if (!semid.empty())
{
client.set_session_id(semid);
client << "<p>new session_id:|" << client.get_session_id() << "|</p>";
client << "<p>new session aaa:|" << client.session["aaa"] << "|</p>";
client << "<p>cookie same:" << client.cookie["same"] << "</p>";
client << "<p>cookie ccbb:" << client.cookie["ccbb"] << "</p>";
client << "<p>session aaa:|" << client.session.to_json() << "|</p>";
}
client.view("cookie/show");
return "";
}
save_session() save session values to disk or memory.
The framework has three built-in methods for handling HTML data transmission and processing, all of which are built-in micro objects of type http::obj_val, The three objects are client.get[] client.post[] client.files[]
URL parameter value transmission, use client.get[] get values, example http://www.xxx.com/aa?username=root&userid=1111 use client.get["username"].to_string() and client.get["userid"].to_int() get the values.
HTML form post data use client.post[] get the values
Example file www/default/testpost.html, The HTML code is as follows.
<form method="post" enctype="multipart/form-data" name="form1" id="form1" action="/tfilepost">
<p>
<label for="textfield">Text Field:</label>
<input type="text" name="textfield" id="textfield">
</p>
<p>
<label for="textfield">Text Field:</label>
<input type="text" name="aa[]" id="textfield1">
</p>
<p>
<label for="textfield">Text Field:</label>
<input type="text" name="aa[]" id="textfield2">
</p>
<p>
<label for="textfield">Text Field:</label>
<input type="text" name="aa[8]" id="textfield3">
</p>
<p>
<label for="textarea">Text Area:</label>
<textarea name="textarea" id="textarea"></textarea>
</p>
<p>
<label for="fileField">File:</label>
<input type="file" name="fileField" id="fileField">
</p>
<p> </p>
<p>
<input type="submit" name="submit" id="submit" value="Submit">
</p>
</form>
Server side receives form data code, Pay attention to the usage of the value of client.post["aa"].
Example file controller/src/testformpost.cpp
//@urlpath(null,tfilepost)
std::string testformmultipart(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
client << "hello world! multipart/form-data ";
try
{
client << "<p>--------------</p>";
client << "<p>";
client << client.post.to_json();
client << "</p>";
client << "<p>textfield:" << client.post["textfield"] << "</p>";
client << "<p>aa[]:" << client.post["aa"][0] << "</p>";
client << "<p>aa[]:" << client.post["aa"][1] << "</p>";
client << "<p>aa[8]:" << client.post["aa"][8] << "</p>";
client << "<p>textarea:" << client.post["textarea"] << "</p>";
client << "<p>fileField:name:" << client.files["fileField"]["name"] << "</p>";
client << "<p>fileField:filename:" << client.files["fileField"]["filename"] << "</p>";
client << "<p>fileField:tempfile:" << client.files["fileField"]["tempfile"] << "</p>";
client << "<p>fileField:size:" << client.files["fileField"]["size"] << "</p>";
client << "<p>fileField:error:" << client.files["fileField"]["error"] << "</p>";
client << "<p>submit:" << client.post["submit"] << "</p>";
client << "<p>--------------</p>";
}
catch (std::exception &e)
{
client << "<p>" << e.what() << "</p>";
return "";
}
catch (const char *e)
{
client << "<p>" << e<< "</p>";
return "";
}
return "";
}
client.files["fileField"]["tempfile"] save to temporary path on the server side, use std::filesystem::rename(client.files["fileField"]["tempfile"].as_string(), newfilepath) function, Save to the correct server side location.
The data imported in the installation environment section.
The paozhu framework comes with built-in ORM objects and utilizes the built-in MySQL protocol library, eliminating the need for third-party libraries and facilitating integration with ORM. It also supports C++ Coroutine mode. The MySQL official client does not support Coroutines.
Config file at conf/orm.conf
[default]
type=main
host=127.0.0.1
port=3306
dbname=hello_world
user=benchmarkdbuser
password=benchmarkdbpass
pretable=
maxpool=5
dbtype=mysql
charset=utf8mb4
#ssl=ON
type=second
host=127.0.0.1
port=3306
dbname=hello_world
user=benchmarkdbuser
password=benchmarkdbpass
pretable=
maxpool=20
dbtype=mysql
charset=utf8mb4
#ssl=ON
type The type is Master-Slave Replication and cache synchronization can be developed in the future. Currently, master-slave is preferred. Host '127.0.0.1' fills in the local address, MySQL must be greater than or equal to version 8 because MySQL 9 does not support old authentication.
pretable It is a table prefix, sometimes only one database permission can store multiple businesses in one database.
dbtype Currently supports MySQL.
SSL can be opened, especially when connecting remotely, enabling encrypted connections. Local SSL connections will not be enabled.
The framework currently supports database to create cpp file conversion, so modifying the database can regenerate ORM files.
After the project is successfully compiled, the following method can be used to generate cpp files.
# ./bin/paozhu_cli
paozhu_cli(1495,0x7ff85ac38fc0) malloc: nano zone abandoned due to inability to reserve vm space.
model ｜ view | viewtocpp | control
Welcome to use cli to manage your MVC files。
(m)model (v)view (f)viewtocpp or (c)control , (j)son ,x or q to exit[input m|v|f|c|j|]:
Enter m to select and generate ORM model file
paozhu_cli(1495,0x7ff85ac38fc0) malloc: nano zone abandoned due to inability to reserve vm space.
model ｜ view | viewtocpp | control
Welcome to use cli to manage your MVC files。
(m)model (v)view (f)viewtocpp or (c)control , (j)son ,x or q to exit[input m|v|f|c|j|]:m
🍄 current path: /Users/hzq/paozhu
1 hello_world
3 paozhu_docs
5 cppcms
select db index:
Enter 1 to select the hello-world database to generate the file.
paozhu_cli(1495,0x7ff85ac38fc0) malloc: nano zone abandoned due to inability to reserve vm space.
model ｜ view | viewtocpp | control
Welcome to use cli to manage your MVC files。
(m)model (v)view (f)viewtocpp or (c)control , (j)son ,x or q to exit[input m|v|f|c|j|]:m
🍄 current path: /Users/hzq/paozhu
1 hello_world
3 paozhu_docs
5 cppcms
select db index:1
show tables;
create fortune table to models 🚗
create table metainfo file: /Users/hzq/paozhu/orm/include/fortune_base.h
create world table to models 🚗
create table metainfo file: /Users/hzq/paozhu/orm/include/world_base.h
(m)model (v)view (f)viewtocpp or (c)control , (j)son ,x or q to exit[input m|v|f|c|j|]:x
Enter x to exit
You can see the generated files in orm directory.
orm
├── include
│ ├── fortune_mysql.h
│ ├── fortune_base.h
│ ├── world_mysql.h
│ └── world_base.h
├── orm.h
│
models
├── Fortune.cpp
├── World.cpp
└── include
├── Fortune.h
└── World.h
The ORM directory is an entity file, consisting of two files: one is the end of the mysql.h for operating database query interfaces, and the other is the end of the data table entity mapping base.h, which is also an interface for operating query results. When updating this directory file, it will be overwritten because it corresponds one-to-one with the database table structure
The models directory is the business code file, and we actually manipulate the files in the models directory. These two files, one cpp and one h, are automatically generated, but they are only generated once and will not be overwritten. In our own project system, we can add business code to them.
For example
auto myworld = orm::World(); models directory World.h
This is the default database, such as the built-in CMS demo auto users = orm::cms::Sysuser();
orm is namespae, Mainly used for ORM business CMS is the orm.conf file database configuration tag and also the generated CMS namespace, which facilitates database isolation and transformation. If it is [default], default can be omitted
Why is it designed in this way? Actually, it is also a unique design of the Paozhu framework compared to other frameworks.
So orm::cms::Sysuser(); Using CMS to isolate the 'sysuser' table entity C++ class will not conflict with tables with the same name in other databases.
For example, orm::erp::Sysuser(); ERP is an ERP database tag, and the name of its database is not important,
It doesn't matter if there is also a sysuser table.
The reason for this design is that it is a business system, a single server program that can have many databases, and their table names can have duplicate names. In the future, there can be different databases, such as MySQL PG, which can be distributed on other machines. They can mix clusters, and individual server programs can also be clustered, because they have built-in read-write separation, and the write servers can be the same, Querying can allocate database servers in a fixed or dynamic manner, and the framework can be suitable for large business systems. Currently, the focus is on integrating a server with a database and business programs.
ORM does not have multi table join queries and requires manual writing of multi table queries. This may result in one or more additional queries at the beginning of the design, mainly due to fast in and fast out, which makes it easy to hit the cache and is in line with the current storage computing separation design. It also performs smoothly under high pressure, and novice programmers will not encounter any problems when writing business code.
The table model has two important member variables
Table Data Structure and Data table entity structure
Reference file orm/include/world_base.h
struct meta{
unsigned int id = 0;
int randomnumber = 0;
} data;
std::vector<world_base::meta> record;
//@urlpath(null,queries)
asio::awaitable<std::string> techempowerqueries(std::shared_ptr<httppeer> peer)
{
peer->type("application/json; charset=UTF-8");
peer->set_header("Date", get_gmttime());
unsigned int get_num = peer->get["queries"].to_int();
if (get_num == 0)
{
get_num = 1;
}
else if (get_num > 500)
{
get_num = 500;
}
auto myworld = orm::World();
myworld.record.reserve(get_num);
myworld.lock_conn();
for (unsigned int i = 0; i < get_num; i++)
{
myworld.wheresql.clear();
unsigned int rd_num = rand_range(1, 10000);
myworld.where("id", rd_num);
co_await myworld.async_fetch_append();
}
myworld.unlock_conn();
peer->output = myworld.to_json();
co_return "";
}
orm::World(); It is an ORM table model with default connections, so there is no need for defaultmyworld.lock_conn(); It's about fixing a connection, because using ORM connection pool method will rotate the use of connections.myworld.record.reserve(get_num); The std::vector reserve。myworld.wheresql.clear(); Recommend using myworld.clearWhere() Method to clear conditions。myworld.where("id", rd_num); Add where。co_await myworld.async_fetch_append(); It is an advanced usage that will attach data to the record object. Generally, co_await myworld.async_fetch(); is used` It will clear the record object and then reassign it.peer->output = myworld.to_json(); Directly output JSON to the browser.
Normal business code runs in a thread, one thread by one connect at a time, which is a classic way.
//@urlpath(admin_islogin,admin/addtopic)
std::string admin_addtopic(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
try
{
auto topicm = orm::cms::Topic();
topicm.where("userid", client.session["userid"].to_int()).asc("parentid").fetch();
client.val["list"].set_array();
obj_val temp;
for (unsigned int i = 0; i < topicm.record.size(); i++)
{
temp["id"] = topicm.record[i].topicid;
temp["parentid"] = topicm.record[i].parentid;
temp["value"] = topicm.record[i].title;
client.val["list"].push(temp);
}
}
catch (std::exception &e)
{
client.val["code"] = 1;
}
peer->view("admin/addtopic");
return "";
}
httppeer &client = peer->get_peer(); Reference connection object
auto topicm = orm::cms::Topic(); Create an ORM model object, with the model source file located in the models directory,Topic is the database table name，If a prefix is set, the framework will automatically remove it,topicm.where add sql where，asc("parentid").fetch() order parentid asc query ，.fetch() send query command。
If you don't want so much data to be paginated and retrieved, especially for network services that require bandwidth or CPU resources，It is best to use it as needed .select("fieldname1,fieldname2") Retrieve field data, use .limit(pos,offset) for range.
Use .page(page, 10, 5) It has been automatically added .limit(pos,offset) to sql string,now only input page number.
auto [bar_min, bar_max, current_page, total_page] = artmodel.page(page, 10, 5);
client.val["pageinfo"].set_object();
client.val["pageinfo"]["min"] = bar_min;
client.val["pageinfo"]["max"] = bar_max;
client.val["pageinfo"]["current"] = current_page;
client.val["pageinfo"]["total"] = total_page;
.page(page, 10, 5) Return four numerical values,bar_min bar_max The range is determined by the input of 5, It consists of two pages of windows, with 10 per page,
ORM where Other Methods
whereLike(fieldname,searchword) It's equivalent to and fieldname '%searchword%' alias
whereAnd It's equivalent to and fieldname = val alias
whereIn iIt's equivalent to and fieldname IN(val) alias
whereLT(const std::string &wq, const std::string &val) It's equivalent to and fieldname < val alias
whereLE(const std::string &wq, const std::string &val) It's equivalent to and fieldname <= val alias
whereOrLT(const std::string &wq, const std::string &val) It's equivalent to OR fieldname < val alias
whereOrLE(const std::string &wq, const std::string &val) It's equivalent to OR fieldname <= val alias
whereBT(const std::string &wq, const std::string &val) It's equivalent to and fieldname > val alias
whereBE(const std::string &wq, const std::string &val) It's equivalent to and fieldname >= val alias
whereOrBT(const std::string &wq, const std::string &val) It's equivalent to OR fieldname > val alias
whereOrBE(const std::string &wq, const std::string &val) It's equivalent to OR fieldname >= val alias
ORM Execute Command
Synchronous mode
fetch() Get multiple results to variable record
fetch_one() Get a result to variable data
save() Insert variable data into database table
update("field1,field2...") Update fields in variable data to the database
remove() Delete data from the table and match it with the where condition
Coroutine mode
async_fetch() Get multiple results to variable record using coroutine method
async_fetch_one() Get a result to variable data through coroutine method
async_save() Insert variable data into database table using coroutine method
async_update("field1,field2...") Update fields in variable data to the database using coroutine method
async_remove() Delete data from the table and match it with the where condition using coroutine method
For more information, please refer to the methods in the file ending with mysql.h in the ORM directory.
Reference documents controller/src/superadmin/supermain.cpp Example of data insertion.
//@urlpath(superadmin_islogin,superadmin/adduserpost)
std::string superadmin_adduserpost(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
unsigned int sid = client.post["sid"].to_int();
auto stinfo = orm::cms::Siteinfo();
stinfo.where("agentid", client.session["superid"].to_int()).whereAnd("sid", sid).fetch_one();
if (stinfo.data.userid == 0)
{
client.goto_url("/superadmin/welcome", 3, "No site information!");
return "";
}
std::string expdate = str_trim(client.post["expectdate"].to_string());
if (!check_isodate(expdate))
{
expdate = get_date("%Y-%m-%d");
}
auto uinfo = orm::cms::Sysuser();
std::string username = client.post["username"].to_string();
std::string password = client.post["password"].to_string();
uinfo.where("name", username).fetch_one();
std::string urlmsg = "/superadmin/adduser?sid=";
urlmsg.append(std::to_string(sid));
if (uinfo.data.adminid > 0)
{
client.goto_url(urlmsg, 3, "The username already exists!");
return "";
}
if (username.size() > 26 || username.size() < 4)
{
client.goto_url(urlmsg, 3, "The username must be between 4 and 20!");
return "";
}
if (password.size() > 20 || password.size() < 6)
{
client.goto_url(urlmsg, 3, "The password must be between 4 and 20!");
return "";
}
uinfo.clear();
uinfo.data.name = username;
uinfo.data.password = md5(password);
uinfo.data.isopen = client.post["isopen"].to_int() == 1 ? 1 : 0;
uinfo.data.companyid = sid;
uinfo.data.textword = password;
uinfo.data.created_at = timeid();
uinfo.data.enddate = strtotime(expdate);
uinfo.data.nickname = client.post["nickname"].to_string();
uinfo.data.mobile = client.post["mobile"].to_string();
uinfo.data.email = client.post["email"].to_string();
auto [effect_num, last_id] = uinfo.save();
if (effect_num == 0)
{
client.goto_url(urlmsg, 3, "Save error!");
return "";
}
urlmsg = "/superadmin/listuser?sid=";
urlmsg.append(std::to_string(sid));
client.goto_url(urlmsg, 3, "User added successfully!");
return "";
}
auto [effect_num, last_id] = uinfo.save(); The method returns a tuple value, where effectod_num represents the number of rows inserted, and last_id represents the self increasing id of the first inserted data.
Reference documents controller/src/admin/topics.cpp Example of data update.
//@urlpath(admin_isloginjson,admin/updatetopicview)
std::string admin_updatetopicview(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
try
{
auto topicm = orm::cms::Topic();
unsigned char isview = client.post["isview"].to_int() > 0 ? 1 : 0;
topicm.setIsview(isview);
topicm.where("userid", client.session["userid"].to_int()).whereAnd("topicid", client.get["id"].to_int());
client.val["code"] = topicm.update("isview");
client.val["msg"] = "ok";
}
catch (std::exception &e)
{
client.val["code"] = 0;
}
client.out_json();
return "";
}
client.val["code"] = topicm.update("isview"); update("filed1,field2...") Indicates the need to update fields
Paozhu, as a C++web framework framework, is the MVC pattern, which now refers to the view function of V
It is the content displayed to the browser, which is actually HTML content, similar to PHP scripting language output. So how do we operate C++? We put C++code in HTML and compile it into a C++. cpp file, which is resident in memory. This is also very fast. Hundreds of view files only require a few MB of memory, and one company project is about 1000 pages. It is estimated to have less than 10MB of memory, and if running in a scripting language, it will require a large amount of I/O to read files. The framework compiles it directly into a program without the need for I/O reading, which is extremely fast.
We are in the view directory
Create a login directory and then create a login.html file in the login directory
view
└── login
└── login.html
Probably like this
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta name="description" content="">
<meta name="author" content="Mark Otto, Jacob Thornton, and Bootstrap contributors">
<meta name="generator" content="Hugo 0.101.0">
<title>Content Management System</title>
<link href="/assets/dist/css/bootstrap.min.css" rel="stylesheet">
</head>
<body class="text-center">
<%c include_sub("home/header",obj); %>
<div class="container text-center">
<div class="row">
<div class="col-3"></div>
<div class="col-6">
<h2 id="horizontal-form">CMS Admin </h2>
<form action="/cms/loginpost" method="post">
<div class="row mb-3">
<label for="username" class="col-sm-2 col-form-label">Username</label>
<div class="col-sm-10">
<input type="text" class="form-control" id="username" name="username" value="admin">
</div>
</div>
<div class="row mb-3">
<label for="password" class="col-sm-2 col-form-label">Password</label>
<div class="col-sm-10">
<input type="password" class="form-control" id="password" name="password" value="123456">
</div>
</div>
<button type="submit" class="btn btn-primary">Sign in</button>
</form>
</div>
</div>
</div>
</body>
</html>
include_sub("home/header",obj);
The description also includes a subpage, and we will create a subpage directory called home in the view
view
├── home
│ └── header.html
└── login
└── login.html
Probably like this
<style>
ul {
list-style-type: none;
margin: 0;
padding: 0;
overflow: hidden;
background-color: #333;
}
li {
float: left;
border-right:1px solid #bbb;
}
li:last-child {
border-right: none;
}
li a {
display: block;
color: white;
text-align: center;
padding: 14px 16px;
text-decoration: none;
}
li a:hover:not(.active) {
background-color: #111;
}
.active {
background-color: #4CAF50;
}
</style>
<ul>
<li><a class="active" href="/">HOME</a></li>
<li><a href="#news">News</a></li>
<li><a href="#contact">Contact</a></li>
<li style="float:right"><a href="#about">About</a></li>
</ul>
<h3> subpage include_sub("home/header")Nav test |<%c echo<<vinfo.get["aa"].to_string(); %>| </h3>
echo<<vinfo.get["aa"].to_string();
This is the view tag C++code included in <%c %>
echo is a buffer, which is assembled together with the external HTML
Convert to C++ code
It's like this
echo<<"Nav test |";
echo<<vinfo.get["aa"].to_string();
echo<<"| </h3>";
You can refer to the viewsrc/view/home/head.cpp code for details.
Run in the root directory of the website
./bin/paozhu_cli
Select 'f' to generate files
./bin/paozhu_cli
./bin/paozhu_cli model ｜ view | viewtocpp | control
🎉 Welcome to use cli to manage your MVC files。
(m)model (v)view (f)viewtocpp or (c)control,x or q to exit[input m|v|f|c|]:f
🍄 current path: /Users/hzq/paozhu
1 [+] view/home/header.html time: 2022-12-12 13:1:41
2 [+] view/login/login.html time: 2022-12-12 13:1:26
please input number to parse to cpp file.
a update all , example: 1 3 4 5 enter key, q or x to exit,r reload
input number:a
We input f and compile it directly with the framework.
Then enter a, and all modified or newly created view files will be converted to cpp files
The file is saved in the viewsrc directory
viewsrc
├── include
│ ├── regviewmethod.hpp
│ └── viewsrc.h
└── view
├── home
│ └── header.cpp
└── login
└── login.cpp
Probably like this
The login.html header.html files have been converted to cpp files
Two files in the include directory
Viewsrc.h is the definition of all view file functions
regviewmethod.hpp is a view registration function that the framework calls to register with the framework
#ifndef __HTTP_VIEWSRC_ALL_METHOD_H
#define __HTTP_VIEWSRC_ALL_METHOD_H
#if defined(_MSC_VER) && (_MSC_VER >= 1200)
#pragma once
#endif // defined(_MSC_VER) && (_MSC_VER >= 1200)
#include<string>
#include<map>
#include<functional>
#include "request.h"
#include "viewso_param.h"
namespace http {
namespace view {
namespace home{
std::string header(const struct view_param &vinfo,http::obj_val &obj);
}
namespace login{
std::string login(const struct view_param &vinfo,http::obj_val &obj);
}
}
}
#endif
You can see that the views are all placed in the namespace, so there is no conflict between the view files in one directory and those in other directories with the same name.
#ifndef __HTTP_REG_VIEW_METHOD_HPP
#define __HTTP_REG_VIEW_METHOD_HPP
#if defined(_MSC_VER) && (_MSC_VER >= 1200)
#pragma once
#endif // defined(_MSC_VER) && (_MSC_VER >= 1200)
#include<string>
#include<map>
#include<functional>
#include "request.h"
#include "viewso_param.h"
#include "viewmethold_reg.h"
#include "viewsrc.h"
namespace http
{
void _initview_method_regto(VIEW_REG &_viewmetholdreg)
{
//create time: Mon, 12 Dec 2022 05:01:53 GMT
_viewmetholdreg.emplace("home/header",http::view::home::header);
_viewmetholdreg.emplace("login/login",http::view::login::login);
}
}
#endif
Two functions are registered on the callback function, which means a view file is a function
Directory name and file name form a registration point
Login/login is a registration point
Controller peer->view("login/login"); calling method.
We have created the view, now let's test it
Follow the introductory chapter of Hello World
controller/srcDirectory create testview.cpp file
Probably like this
#include <chrono>
#include <thread>
#include "httppeer.h"
#include "testview.h"
namespace http
{
//@urlpath(null,testview)
std::string testloginview(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->getpeer();
client << " view test ";
// client << client.gethosturl();
// client<<"<p><a href=\""<<client.gethosturl()<<"/showcookie\">show</a></p>";
peer->view("login/login");
return "";
}
}
Map testview to testloginview function, the two names can be different.
http://localhost/testview You can access our testloginview function
Everything is ready, let's start compiling
Return to the project root directory and enter the build directory
cmake ..
make
Then go back to the root directory or open a new Terminal Command
cd ..
./bin/paozhu #In Linux use: sudo ./bin/paozhu
Open with a browser
http://localhost/testview
You can see the content of the view now
The framework has built-in structure and JSON conversion functions, reducing the workload of writing code
The framework provides two function operations json_encode json_decode that can be on structures or class objects、
std::string json_encode(obj)
std::string json_encode(std::vector<obj>)
Encoding objects as JSON supports arrays or objects
int json_decode(obj,json_string)
int json_decode(std::vector<obj>,json_string)
Decoding JSON strings to objects
common/json_reflect_headers.h It is the header file referenced by all reflection objects
This file was generated bypaozhu_cli
You need to create the corresponding type file in the libs directory, and then run paozhu_cli to generate it. The framework directly analyzes the type file to generate the corresponding code. Currently, it supports commonly used structures, which can be nested in multiple layers or reference structures that are also reflective.
Create test file libs/department/department_type.h
#ifndef LIBS_DEPARTMENT_TYPE_H
#define LIBS_DEPARTMENT_TYPE_H
#include <iostream>
#include <string>
#include <vector>
#include <sstream>
#include <sstream>
#include "unicode.h"
namespace psy
{
//@reflect json to_json from_json
struct department_outjson_t
{
unsigned int id = 0;
unsigned int key = 0;
unsigned int value = 0;
unsigned int parentid = 0;
unsigned int bianzhi_num = 0;
bool isopen = true;
bool _is_use = false;
std::string title;
std::string desc;
std::vector<department_outjson_t> children;
std::string to_json(std::string aa="",int offset=0);
unsigned int from_json()
{
//**aaaaa
/*
aaaa [][}]
*/
char b='}';
if(b=='{')
{
std::cout<<"assss}sss}{ssss}sss}"<<std::endl;
}
return 0;
}
};
//@reflect json to_json from_json
struct department_listoutjson_t
{
unsigned int code=0;
struct department_tables
{
std::vector<department_outjson_t> list;
unsigned int total=0;
} data;
std::vector<std::vector<std::string>> names;
};
} // namespace psy
#endif
//@reflect json It is a framework analysis source code extraction flag, which must be a struct or class starting with this, otherwise it will not work.
We create two structures, with one nested within the other, to facilitate business code operations. to_json from_json is a testing framework that analyzes whether the source code correctly skips these functions
Variables starting with an underscore will not be output to a JSON string
json_encode and json_decode
sudo ./bin/paozhu_cli
./bin/paozhu_cli model | view | viewtocpp | control
🎉 Welcome to use cli to manage your MVC files。
(m)model (v)view (f)viewtocpp or (c)control , (j)son ,x or q to exit[input m|v|f|c|j|]:
Input j
You can see the generated file now
1 /Users/hzq/paozhu/libs/department/department_type_jsonreflect.cpp
(m)model (v)view (f)viewtocpp or (c)control , (j)son ,x or q to exit[input m|v|f|c|j|]:
Reference controller/src/testjsonreflect.cpp
#include <chrono>
#include <thread>
#include "httppeer.h"
#include "testjsonreflect.h"
#include "json_reflect_headers.h"
#include "array_to_tree.h"
namespace http
{
//@urlpath(null,testjsonreflect)
std::string testjsonreflect(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->getpeer();
// client << " json reflect 🧨 Paozhu c++ web framework ";
psy::department_outjson_t deps_json_one;
std::vector<psy::department_outjson_t> depsjsonlist;
deps_json_one.id = 38;
deps_json_one.key = 38;
deps_json_one.value = 38;
deps_json_one.parentid = 0;
deps_json_one.isopen = true;
deps_json_one.title = "First Item";
deps_json_one.desc = "First Item";
deps_json_one.bianzhi_num = 12;
depsjsonlist.push_back(deps_json_one);
deps_json_one.id = 48;
deps_json_one.key = 48;
deps_json_one.value = 48;
deps_json_one.parentid = 0;
deps_json_one.isopen = true;
deps_json_one.title = "Second Item";
deps_json_one.desc = "Second memo";
deps_json_one.bianzhi_num = 16;
depsjsonlist.push_back(deps_json_one);
deps_json_one.id = 58;
deps_json_one.key = 58;
deps_json_one.value = 58;
deps_json_one.parentid = 48;
deps_json_one.isopen = true;
deps_json_one.title = "Three Item";
deps_json_one.desc = "Three memo";
deps_json_one.bianzhi_num = 22;
depsjsonlist.push_back(deps_json_one);
psy::department_listoutjson_t depout_data;
array_to_tree<psy::department_outjson_t>(depout_data.data.list, depsjsonlist);
depout_data.code = 0;
// depout_data.data.list=depsjsonlist; //has array_to_tree
depout_data.data.total = 0;
std::string jsondep_str = psy::json_encode(depout_data);
psy::department_listoutjson_t new_depout_data;
psy::json_decode(new_depout_data, jsondep_str);
std::cout << "data.list:" << depout_data.data.list.size() << std::endl;
std::cout << "data.list:" << new_depout_data.data.list.size() << std::endl;
std::string jsondep_str_temp = psy::json_encode(new_depout_data);
client << jsondep_str_temp;
client.out_jsontype();
return "";
}
} // namespace http
We first convert the array into a tree, and then encode the test object into a JSON string
std::string jsondep_str = psy::json_encode(depout_data);
Then decode the jsondep_str variable to the new object
psy::json_decode(new_depout_data, jsondep_str);
And then we'll do it again std::string jsondep_str_temp = psy::json_encode(new_depout_data);
Then output jsondep_str_temp, See if it meets our expectations
Example file controller/src/testcowaitclient.cpp
This is the synchronous usage method
//@urlpath(null,testcowaitclient5)
std::string testhttpclient_cowait_post(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
client << "hello world! test testhttpclient_cowait_body";
std::shared_ptr<http::client> a = std::make_shared<http::client>();
a->get("http://www.xxxxxx.net/");
a->add_header("Connection", "keep-alive");
a->send();
if (a->get_status() == 200)
{
// client << a->getBody();
std::cout << "http://www.xxxxxx.net/login.php" << std::endl;
a->post("http://www.xxxxxx.net/login.php");
a->add_header("Connection", "keep-alive");
a->data["user_name"] = "admin";
a->data["user_pass"] = "123456";
a->data["action"] = "login";
a->send();
if (a->get_status() == 200)
{
//client << a->getBody();
std::cout << "http://www.xxxxxx.net/main.php" << std::endl;
a->get("http://www.xxxxxx.net/main.php");
a->requst_clear();
a->add_cookie("PHPSESSID", a->state.cookie["PHPSESSID"]);
a->add_header("Connection", "Close");
a->send();
if (a->get_status() == 200)
{
std::cout << a->get_body() << std::endl;
//client << a->getBody();
}
}
}
else
{
client << a->error_msg;
client << "+++++not content++++";
}
return "";
}
This is the Coroutine usage method
//@urlpath(null,testcowaitclient22)
asio::awaitable<std::string> testhttpclient22_cowait_body(std::shared_ptr<httppeer> peer)
{
httppeer &client = peer->get_peer();
client << "hello world! co_await test localhost http://127.0.0.1addpost.html ";
std::shared_ptr<http::client> a = std::make_shared<http::client>();
a->get("http://127.0.0.1/addpost.html");
co_await a->async_send();
if (a->get_status() == 200)
{
client << a->get_body();
}
else
{
client << "+++++not content++++";
}
co_return "";
}
the Coroutine use async_ prefix.
The framework has enabled ASAN, and if it is detected by ASAN during debugging, it will indicate which line is problematic.
The production environment is difficult to see due to the lack of output reasons.
The compilation parameters for the production environment.
mkdir build
cd build
cmake .. -DCMAKE_BUILD_TYPE=Release #OR cmake .. -DCMAKE_BUILD_TYPE=Release -DENABLE_GD=ON
make -j8
Need to install gdb and coredumpctl
The framework will be checked every minute, at log directory have a restart_server file
If we need to retrieve a sample of the process stack, we can touch log/start_Server so that the process can restart
have access to pstree -aup view process status
If the sample has already been taken, I need to analyze the process stack situation
view coredump info
coredumpctl list
coredumpctl info
Looking directly online 1524 pid, That's what I saw beforepstree up
coredumpctl gdb 1524bt can be view it
After entering, we can save it and view it in the currentthread_info.txtfile.
set logging file thread_info.txt
set logging enabled on
thread apply all bt
set logging enabled off
The core coredump file is usually located in
/var/lib/systemd/coredumpIf compiled using -g, you will see something similar to the following, usually displaying what various threads are currently doing
Thread 17 (Thread 0x7fd9f16b9640 (LWP 10658)):
#0 http::hash_objkey (key="data:image/php;base64,PD9waHAgJE8wME9PMD11cmxkZWNvZGUoIiU3OCUzNCU2My"...) at /www/paozhu/vendor/httpserver/src/request.cpp:54
#1 0x000000000060032f in http::obj_val::operator[] (this=0x7fd9cc01e520, key="data:image/php;base64,PD9waHAgJE8wME9PMD11cmxkZWNvZGUoIiU3OCUzNCU2MyU2RiUyRiU3MCUzOSU3OSU3MSU2RSU2NCUyRCU2QyU3MiU2QiU2NCU2NyU1RiU"...) at /www/paozhu/vendor/httpserver/src/request.cpp:315
#2 0x00000000005a990c in http::httpparse::procssxformurlencoded (this=0x7fd9b400d370) at /www/paozhu/vendor/httpserver/src/http_parse.cpp:389
#3 0x00000000005b1bfa in http::httpparse::process (this=0x7fd9b400d370, buffer=0x2e223c0 "POST /public/static/lib/webuploader/0.1.5/server/preview.php HTTP/1.1\r\nHost: www. paozhu.org\r\nUser-Agent: Mozilla/5.0 AppleWebKit/601.1.46 (KHTML, like Gecko) Version/9.0\r\nAccept-Encoding: gzip, deflate"..., buffersize=964) at /www/paozhu/vendor/httpserver/src/http_parse.cpp:2823