oracle我如何在过程/包中找到开始的行号

Posted 2023-03-28

【中文标题】oracle我如何在过程/包中找到开始的行号

【英文标题】：oracle how can i find the line number for the begin in a procedure / package

【发布时间】：2014-06-02 11:48:20

【问题描述】：

我想在每个程序、函数和打包程序的开始后添加一些代码，我有成千上万的程序、程序包等，因此手动执行此操作将花费数周的时间。

我可以搜索 all_source，但在我编写自己的代码以查找“开始”的行号之前，是否有人有任何分析代码，因为可能存在我需要忽略的子例程或匿名块

例如来自用户源

package body my_pck is
  procedure proc1 
  is
    v_var varchar2(10);
    procedure sub_proc
    is
    begin
      some code ....
    end;
  begin
< I want to insert code here>
    some code ....

    begin
      anon block ....
    end;
  end;
  procedure proc2 
  is
    v_var varchar2(10);
    procedure sub_proc_something_else
    is
    begin
      some code ....
    end;
    procedure sub_proc_another
    is
    begin
      some code ....
    end;
  begin
< I want to insert code here as well>
    some code ....
  end;
end;

所以我认为我需要一个例程来匹配开始/结束以找到添加我的代码的位置

谢谢 罗伯特

【参考方案1】：

你肯定需要一个解析器来完成这个任务。我为Vorax 创建了一个plsql 解析器，如果你愿意，可以使用它。它不是一个成熟的解析器，但足够聪明，可以弄清楚 plsql 代码的结构。

从 github 获取 vorax4 项目。

cd C:\Projects
git clone https://github.com/talek/vorax4
gem install vorax

---

编写一个简单的 ruby​​ 代码来处理您的场景：

$LOAD_PATH << 'C:/Projects/vorax4/vorax/ruby/lib/'

require 'vorax.rb'
include Vorax

# You need to get this from your database
package_source =<<EOF

package body my_pck is
  procedure proc1 
  is
    v_var varchar2(10);
    procedure sub_proc
    is
    begin
      some code ....
    end;
  begin
    some code ....

    begin
      anon block ....
    end;
  end;
  procedure proc2 
  is
    v_var varchar2(10);
    procedure sub_proc_something_else
    is
    begin
      some code ....
    end;
    procedure sub_proc_another
    is
    begin
      some code ....
    end;
  begin
    some code ....
  end;
end;

EOF

structure = Parser::PlsqlStructure.new(package_source)

# just print the structure of the package
puts structure.dump

# the code we want to insert
offset = 0
INSERT_CODE = "\n--my code\n"
INSERT_CODE_LEN = INSERT_CODE.length

# loop into the plsql structure
structure.regions.each do |child|

  region = child.content

  # only if it's a procedure/function declared into the body of the package
  if region && child.level == 2 && region.instance_of?(Parser::SubprogRegion)
    pos_to_insert = offset + region.body_start_pos + 'begin'.length

    # insert the code after the BEGIN clause
    package_source.insert(pos_to_insert, INSERT_CODE)

    # adjust the offset
    offset += INSERT_CODE_LEN + 1
  end

end

# print the new version of the package
puts package_source

输出是：

[Level: 0]
  [Level: 1] PackageBodyRegion: :start_pos=>2, :end_pos=>467, :name=>"my_pck", :name_pos=>15, :signature_end_pos=>24, :declare_end_pos=>462
    [Level: 2] SubprogRegion: :start_pos=>27, :end_pos=>217, :name=>"proc1", :name_pos=>37, :body_start_pos=>145
      [Level: 3] SubprogRegion: :start_pos=>77, :end_pos=>141, :name=>"sub_proc", :name_pos=>87, :body_start_pos=>107
      [Level: 3] AnonymousRegion: :start_pos=>175, :end_pos=>210
    [Level: 2] SubprogRegion: :start_pos=>221, :end_pos=>462, :name=>"proc2", :name_pos=>231, :body_start_pos=>432
      [Level: 3] SubprogRegion: :start_pos=>271, :end_pos=>350, :name=>"sub_proc_something_else", :name_pos=>281, :body_start_pos=>316
      [Level: 3] SubprogRegion: :start_pos=>356, :end_pos=>428, :name=>"sub_proc_another", :name_pos=>366, :body_start_pos=>394

package body my_pck is
  procedure proc1
  is
    v_var varchar2(10);
    procedure sub_proc
    is
    begin
      some code ....
    end;
  begin

--my code
    some code ....

    begin
      anon block ....
    end;
  end;
  procedure proc2
  is
    v_var varchar2(10);
    procedure sub_proc_something_else
    is
    begin
      some code ....
    end;
    procedure sub_proc_another
    is
    begin
      some code ....
    end;
  begin

--my code
   some code ....
  end;
end;

其他解析器选项，虽然我没有使用过，但可能是：

ANTLR 语法：https://github.com/porcelli/plsql-parser JAVACC 语法：https://java.net/downloads/javacc/contrib/grammars/PlSql.jj

【参考方案2】：

我认为这将很难（但并非不可能）做到。首先，您需要从数据库中获取包。我有一些代码可以做到这一点：

DECLARE
  package_name user_source.name%type;
  last_package_name user_source.name%type;
  package_body CLOB;
BEGIN
  FOR package_cur IN
  ( SELECT * FROM user_source WHERE type = 'PACKAGE BODY' ORDER BY name, line
  )
  LOOP
    package_name        := package_cur.name;
    IF last_package_name = package_name OR last_package_name IS NULL THEN
      package_body      := package_body || package_cur.text;
    ELSE
      --=======================================================
      -- whole package is now in package_body and can be altered
      --=======================================================
      insert into tmp_table values (package_body);

      -- move to next package
      package_body := package_cur.text;
    END IF;
    last_package_name := package_name;
  END LOOP;
  -- processing very last package
  insert into tmp_table values (package_body);
END;

下一个障碍是代码可以以多种方式进行概述，据我所知，整个包可以写在一行中。您必须找出一种或另一种方法来首先从包中提取所有过程，然后绑定适当的开始关键字。

【讨论】：

我应该什么时候投反对票？每当您遇到极其草率、不费吹灰之力的帖子或明显不正确且可能危险地不正确的答案时，请使用您的反对票。

【参考方案3】：

使用 $$PLSQL_LINE。另请参见 $$PLSQL_UNIT。

BEGIN
  DBMS_OUTPUT.PUT_LINE ( $$PLSQL_UNIT || '.' || $$PLSQL_LINE || ': hello ' );
END;
/

【讨论】：

谢谢，但这取决于手动将其添加到代码中...我想自动化它...我应该添加我将提取源代码到文本文件，添加代码行然后重新编译回数据库....我现在已经编写了一个例程来做到这一点，我只是在完善具有子例程的逻辑，而不是把我的代码放在那里！