上回书,我们说到飞天玉虎蒋伯芳来到蜈蚣岭,不是,重来,上回咱们说到可以在Erlang Shell里面手工构造,加载并调用一个模块.在那个demo里面,我把多个Form单独生成出来,最后放在一起做compile:forms,是不是可以简单点?我们先看完整的module代码,erl_scan:string之后是什么样子的:

erl_syntax

Eshell V5.10.2 (abort with ^G)

1> Code = "-module(t).\n-export([say/0]).\n\nsay() -> \"Hello world!!\". ".

"-module(t).\n-export([say/0]).\n\nsay() -> \"Hello world!!\". "

2> {ok,Tokens,_} =erl_scan:string(Code).

{ok,[{'-',1},

{atom,1,module},

{'(',1},

{atom,1,t},

{')',1},

{dot,1},

{'-',2},

{atom,2,export},

{'(',2},

{'[',2},

{atom,2,say},

{'/',2},

{integer,2,0},

{']',2},

{')',2},

{dot,2},

{atom,4,say},

{'(',4},

{')',4},

{'->',4},

{string,4,"Hello world!!"},

{dot,4}],

4}

可以看到上面的list里面包含了若干Form,erl_scan可以逐行解析代码:

4> erl_scan:tokens([],Code,1).

{done,{ok,[{'-',1},

{atom,1,module},

{'(',1},

{atom,1,t},

{')',1},

{dot,1}],

2},

"-export([say/0]).\n\nsay() -> \"Hello world!!\". "}

当然还有更简单的方式,dynamic_compile 项目把这些事情都做了,还考虑了更为复杂的情况,比如头文件,宏,注释,record之类的, https://github.com/JacobVorreuter/dynamic_compile/blob/master/src/dynamic_compile.erl 下面就是dynamic_compile使用的一个例子:

Eshell V5.10.2 (abort with ^G)

1> Code = "-module(t).\n-export([say/0]).\n\nsay() -> \"Hello world!!\". ".

"-module(t).\n-export([say/0]).\n\nsay() -> \"Hello world!!\". "

2> dynamic_compile:load_from_string(Code).

{module,t}

3> t:say().

"Hello world!!"

4>

上面拼字符串的方法看起来丑?你可以选择erl_syntax,下面我们用比较"优雅"的方法去创建t模块:

Eshell V5.10.2 (abort with ^G)

1> M = erl_syntax:attribute(erl_syntax:atom(module),[erl_syntax:atom(t)]).

{tree,attribute,

{attr,0,[],none},

{attribute,{tree,atom,{attr,0,[],none},module},

[{tree,atom,{attr,0,[],none},t}]}}

2> MF = erl_syntax:revert(M).

{attribute,0,module,t}

3> E = erl_syntax:attribute(erl_syntax:atom(export),[erl_syntax:list([erl_syntax:arity_qualifier(erl_syntax:atom(say),erl_syntax:integer(0))])]).

{tree,attribute,

{attr,0,[],none},

{attribute,

{tree,atom,{attr,0,[],none},export},

[{tree,list,

{attr,0,[],none},

{list,

[{tree,arity_qualifier,

{attr,0,[],none},

{arity_qualifier,

{tree,atom,{attr,0,[],none},say},

{tree,integer,{attr,0,[],none},0}}}],

none}}]}}

4> ExportForm = erl_syntax:revert(E).

{attribute,0,export,[{say,0}]}

5>

5> Clause= erl_syntax:clause([],[],[erl_syntax:atom(hello_world)]).

{tree,clause,

{attr,0,[],none},

{clause,[],none,[{tree,atom,{attr,0,[],none},hello_world}]}}

6>

6> Function = erl_syntax:function(erl_syntax:atom(say),[Clause]).

{tree,function,

{attr,0,[],none},

{func,{tree,atom,{attr,0,[],none},say},

[{tree,clause,

{attr,0,[],none},

{clause,[],none,

[{tree,atom,{attr,0,[],none},hello_world}]}}]}}

7> FunctionForm = erl_syntax:revert(Function).

{function,0,say,0,[{clause,0,[],[],[{atom,0,hello_world}]}]}

8> {ok, Mod, Bin1} = compile:forms([MF,ExportForm, FunctionForm]).

{ok,t,

<<70,79,82,49,0,0,1,208,66,69,65,77,65,116,111,109,0,0,0,

57,0,0,0,6,1,116,...>>}

9> t:say().

** exception error: undefined function t:say/0

10> code:load_binary(Mod, [], Bin1).

{module,t}

11> t:say().

hello_world

12>

Erlang Shared Data using mochiglobal 里面mochiglobal 就是用这种方法实现的,不过,不过,又会有人提意见了,这编写难度也太大了,能折中一下吗?好吧,请出下一位嘉宾smerl

我曾经介绍过开源项目smerl,其定位就是Simple Metaprogramming for Erlang, 我们可以从这份代码里面学到erl_scan erl_parse erl_eval更灵活的应用,项目地址:http://code.google.com/p/smerl/

test_smerl() ->

M1 = smerl:new(foo),

{ok, M2} = smerl:add_func(M1, "bar() -> 1 + 1."),

smerl:compile(M2),

foo:bar(), % returns 2``

smerl:has_func(M2, bar, 0). % returns true

parse_transform

在上篇文章里面,我需要反复生成 Abstract Format,开始手工搞了一下,后来不胜其烦就使用了一下parse_transform.这东西是干什么用的呢?

{parse_transform,Module}
Causes the parse transformation function Module:parse_transform/2 to be applied to the parsed code before the code is checked for errors.

对,它就是允许你在检查错误之前插入自己的逻辑,把那些"奇怪的东西"变成正常的东西,当然你可以选择什么都不做,仅仅打印,看代码:

-module(print_form).

-export([parse_transform/2]).

parse_transform(Forms, _Options) ->

io:format("forms: ~p~n", [Forms]),

Forms.

下面,我们写一个简单的模块a.erl,然后编译它,看输出:

[root@nimbus demo]# cat a.erl

-module(a).

-compile({parse_transform,print_form}).

-export([test/0]).

test()->

"hello world!".

[root@nimbus demo]# erlc -o . -pa . a.erl

forms: [{attribute,1,file,{"a.erl",1}},

{attribute,1,module,a},

{attribute,3,export,[{test,0}]},

{function,5,test,0,[{clause,5,[],[],[{string,6,"hello world!"}]}]},

{eof,10}]

好吧,感冒,鼻子堵得难受,先到这里吧

参考资料:

[0] erl_syntax http://erlang.org/doc/man/erl_syntax.html

[1] Erlang: How to Create and Compile Module in Run-time

http://vas.io/blog/2013/03/23/erlang-how-to-create-and-compile-module-in-run-time/

[2] Hacking-Erlang http://zh.scribd.com/doc/22451864/Hacking-Erlang

最后小图一张,最早在山东卫视凌晨剧场看过第一季,现在已经14季了, 老员工只有Nicky ,Sara ,Jim了:

erl_syntax

Eshell V5.10.2 (abort with ^G)

1> Code = "-module(t).\n-export([say/0]).\n\nsay() -> \"Hello world!!\". ".

"-module(t).\n-export([say/0]).\n\nsay() -> \"Hello world!!\". "

2> {ok,Tokens,_} =erl_scan:string(Code).

{ok,[{'-',1},

{atom,1,module},

{'(',1},

{atom,1,t},

{')',1},

{dot,1},

{'-',2},

{atom,2,export},

{'(',2},

{'[',2},

{atom,2,say},

{'/',2},

{integer,2,0},

{']',2},

{')',2},

{dot,2},

{atom,4,say},

{'(',4},

{')',4},

{'->',4},

{string,4,"Hello world!!"},

{dot,4}],

4}

可以看到上面的list里面包含了若干Form,erl_scan可以逐行解析代码:

4> erl_scan:tokens([],Code,1).

{done,{ok,[{'-',1},

{atom,1,module},

{'(',1},

{atom,1,t},

{')',1},

{dot,1}],

2},

"-export([say/0]).\n\nsay() -> \"Hello world!!\". "}

Eshell V5.10.2 (abort with ^G)

1> Code = "-module(t).\n-export([say/0]).\n\nsay() -> \"Hello world!!\". ".

"-module(t).\n-export([say/0]).\n\nsay() -> \"Hello world!!\". "

2> dynamic_compile:load_from_string(Code).

{module,t}

3> t:say().

"Hello world!!"

4>

上面拼字符串的方法看起来丑?你可以选择erl_syntax,下面我们用比较"优雅"的方法去创建t模块:

Eshell V5.10.2 (abort with ^G)

1> M = erl_syntax:attribute(erl_syntax:atom(module),[erl_syntax:atom(t)]).

{tree,attribute,

{attr,0,[],none},

{attribute,{tree,atom,{attr,0,[],none},module},

[{tree,atom,{attr,0,[],none},t}]}}

2> MF = erl_syntax:revert(M).

{attribute,0,module,t}

3> E = erl_syntax:attribute(erl_syntax:atom(export),[erl_syntax:list([erl_syntax:arity_qualifier(erl_syntax:atom(say),erl_syntax:integer(0))])]).

{tree,attribute,

{attr,0,[],none},

{attribute,

{tree,atom,{attr,0,[],none},export},

[{tree,list,

{attr,0,[],none},

{list,

[{tree,arity_qualifier,

{attr,0,[],none},

{arity_qualifier,

{tree,atom,{attr,0,[],none},say},

{tree,integer,{attr,0,[],none},0}}}],

none}}]}}

4> ExportForm = erl_syntax:revert(E).

{attribute,0,export,[{say,0}]}

5>

5> Clause= erl_syntax:clause([],[],[erl_syntax:atom(hello_world)]).

{tree,clause,

{attr,0,[],none},

{clause,[],none,[{tree,atom,{attr,0,[],none},hello_world}]}}

6>

6> Function = erl_syntax:function(erl_syntax:atom(say),[Clause]).

{tree,function,

{attr,0,[],none},

{func,{tree,atom,{attr,0,[],none},say},

[{tree,clause,

{attr,0,[],none},

{clause,[],none,

[{tree,atom,{attr,0,[],none},hello_world}]}}]}}

7> FunctionForm = erl_syntax:revert(Function).

{function,0,say,0,[{clause,0,[],[],[{atom,0,hello_world}]}]}

8> {ok, Mod, Bin1} = compile:forms([MF,ExportForm, FunctionForm]).

{ok,t,

<<70,79,82,49,0,0,1,208,66,69,65,77,65,116,111,109,0,0,0,

57,0,0,0,6,1,116,...>>}

9> t:say().

** exception error: undefined function t:say/0

10> code:load_binary(Mod, [], Bin1).

{module,t}

11> t:say().

hello_world

12>

test_smerl() ->

M1 = smerl:new(foo),

{ok, M2} = smerl:add_func(M1, "bar() -> 1 + 1."),

smerl:compile(M2),

foo:bar(), % returns 2``

smerl:has_func(M2, bar, 0). % returns true

parse_transform

在上篇文章里面,我需要反复生成 Abstract Format,开始手工搞了一下,后来不胜其烦就使用了一下parse_transform.这东西是干什么用的呢?

{parse_transform,Module}
Causes the parse transformation function Module:parse_transform/2 to be applied to the parsed code before the code is checked for errors.

对,它就是允许你在检查错误之前插入自己的逻辑,把那些"奇怪的东西"变成正常的东西,当然你可以选择什么都不做,仅仅打印,看代码:

-module(print_form).

-export([parse_transform/2]).

parse_transform(Forms, _Options) ->

io:format("forms: ~p~n", [Forms]),

Forms.

下面,我们写一个简单的模块a.erl,然后编译它,看输出:

[root@nimbus demo]# cat a.erl

-module(a).

-compile({parse_transform,print_form}).

-export([test/0]).

test()->

"hello world!".

[root@nimbus demo]# erlc -o . -pa . a.erl

forms: [{attribute,1,file,{"a.erl",1}},

{attribute,1,module,a},

{attribute,3,export,[{test,0}]},

{function,5,test,0,[{clause,5,[],[],[{string,6,"hello world!"}]}]},

{eof,10}]

好吧,感冒,鼻子堵得难受,先到这里吧

参考资料:

[0] erl_syntax http://erlang.org/doc/man/erl_syntax.html

[1] Erlang: How to Create and Compile Module in Run-time

http://vas.io/blog/2013/03/23/erlang-how-to-create-and-compile-module-in-run-time/

[2] Hacking-Erlang http://zh.scribd.com/doc/22451864/Hacking-Erlang

最后小图一张,最早在山东卫视凌晨剧场看过第一季,现在已经14季了, 老员工只有Nicky ,Sara ,Jim了:

[Erlang 0111] Erlang Abstract Format , Part 2相关推荐

[Erlang 0034] Erlang iolist
问题的缘起是芒果在使用mochiweb的过程中遇到一个异常,在google的讨论组中我找到了同样的问题: =ERROR REPORT==== 7-Apr-2011::18:58:22 === &quo ...
[Erlang危机]Erlang In Danger 序言（必读）
原创文章,转载请注明出处:服务器非业余研究http://blog.csdn.net/erlib 作者Sunface 联系邮箱:cto@188.com 这本新书是Learn You Some Erlan ...
[Erlang 0057] Erlang 排错利器: Erlang Crash Dump Viewer
Erlang Crash Dump Viewer真的是排错的天兵神器,还记得我们之前曾经讨论过[Erlang 0013]抓取Erlang进程运行时信息 [Erlang 0012]Erlang Proc ...
[Erlang 0014]Erlang垃圾回收机制
前面的Erlang杂记中我们简单提到过Erlang的垃圾回收机制:1.以进程为单位进行垃圾回收 2.ETS和原子不参与垃圾回收.今天我们继续这一话题,关注更多关于细节. 在Erlang的官方文档中,关 ...
解析Erlang日志组件lager的lager_transform模块
为什么80%的码农都做不了架构师?>>> 使用 lager 的时候,在编译应用的时候,需要加入选项 {parse_transform, lager_transform} erl ...
[Erlang 0079] RabbitMQ 初探
最近在项目中实践RabbitMQ,比较幸运现在除了官方网站,还有一本非常棒的书可以读:RabbitMQ in Action;这本书目前还没有中文版或者影印版,但是从网上很容易找到PDF版本和epub ...
30 分钟学 Erlang
30 分钟学 Erlang (一) Shawn_xiaoyu https://www.jianshu.com/p/b45eb9314d1e 本文写给谁看的? 那些已经有过至少一门编程语言基础,并且需要 ...
ERLANG日期与时间
在开发过程中,有两个概念是和地区区域相关的:字符编码和时间;编码和时间的规范演变过程中有文化的冲突有历史的遗留,是软件开发中充满人文气息的一角;关于字符编码我之前整理过一篇文章, [Erlang 00 ...
Erlang安装 /开启关闭 /配置开发环境 /运行程序
目录: 1. Erlang安装 2. 开启和停止Erlang shell 3. 配置开发环境 4. 运行程序的几种方法 1. Erlang安装 (1). 二进制发布版(此处不做介绍) (2). 从源代 ...

[Erlang 0111] Erlang Abstract Format , Part 2

erl_syntax

parse_transform

erl_syntax

parse_transform

[Erlang 0111] Erlang Abstract Format , Part 2相关推荐

最新文章

热门文章