添加關鍵字提示

Question

我正在尋找編寫一個PowerShell腳本，該腳本將遍歷SQL腳本文件的內容，查找包含TABLE提示的行，其中沒有WITH語句。

例子：

• SELECT * FROM dbo.Table (NOLOCK)
• SELECT * FROM dbo.Table (INDEX=PK_Table)
• SELECT * FROM dbo.Table (NOLOCK,INDEX=PK_Table)
• SELECT * FROM dbo.Table (INDEX=PK_Table,NOLOCK)

我想編寫一個PowerShell腳本看起來對於那些提示並在TABLE HINT之前添加WITH。

所以它們看起來像下面這樣：

• SELECT * FROM dbo.Table WITH(NOLOCK)
• SELECT * FROM dbo.Table WITH(INDEX=PK_Table)
• SELECT * FROM dbo.Table WITH(NOLOCK,INDEX=PK_Table)
• SELECT * FROM dbo.Table WITH(INDEX=PK_Table,NOLOCK)

到目前爲止，我有以下幾點：

foreach ($str in Get-Content "d:\script.sql") { 
    if ([regex]::IsMatch($str, "\({0,1}NOLOCK\)|\(,NOLOCK\{0,1}\)","IgnoreCase")) { 
     $strReplace = [regex]::Replace($str, "\({0,1}NOLOCK\)|\(,NOLOCK\{0,1}\)", "WITH $1") 
     write $strReplace 
    } 
} 

但是，它不插入WITH到行中並保持匹配和行。

這也將不得不處理腳本如下所示：

SELECT Table1.*, Table2.* FROM dbo.Table1 Table1 (NOLOCK) 
INNER JOIN dbo.Table2 Table2 (INDEX=PK_Table2, NOLOCK) 
ON (Table2.Id = Table1.Id) 

下面的建議

：我用下面的：

$expr = "(?=\([^)]*(?=(NOLOCK|INDEX=|INDEX\())[^)]*\))(?<!WITH\s+)" 
 

 
if ([regex]::IsMatch($str, "(?=\([^)]*(?=(NOLOCK|INDEX=|INDEX\())[^)]*\))","IgnoreCase")) { 
 
     $replaced = $str -replace $expr,"WITH "

Answer 1

以下expession會工作，至少在你的輸入樣本中。沒有保證，這將

• 匹配所有的事情應該
• 比賽沒有任何的東西不應該
• 沒有造成嚴重破壞你的SQL代碼

您需要對所有三種情況進行非常仔細的測試也想到邊緣情況。

$sample = " 
SELECT * FROM dbo.Table (NOLOCK) 
SELECT * FROM dbo.Table (INDEX=PK_Table) 
SELECT * FROM dbo.Table (NOLOCK,INDEX=PK_Table) 
SELECT * FROM dbo.Table (INDEX=PK_Table,NOLOCK) 


SELECT * FROM dbo.Table WITH (NOLOCK) 
SELECT * FROM dbo.Table WITH (INDEX=PK_Table) 
SELECT * FROM dbo.Table WITH (NOLOCK,INDEX=PK_Table) 
SELECT * FROM dbo.Table WITH (INDEX=PK_Table,NOLOCK) 
" 

$expr = "(?=\([^()]*(?=\b(?:NOLOCK|INDEX)\b)[^()]*\))(?<!WITH\s+)" 

$sample -replace $expr,"WITH " 

結果：

 
SELECT * FROM dbo.Table WITH (NOLOCK) 
SELECT * FROM dbo.Table WITH (INDEX=PK_Table) 
SELECT * FROM dbo.Table WITH (NOLOCK,INDEX=PK_Table) 
SELECT * FROM dbo.Table WITH (INDEX=PK_Table,NOLOCK) 


SELECT * FROM dbo.Table WITH (NOLOCK) 
SELECT * FROM dbo.Table WITH (INDEX=PK_Table) 
SELECT * FROM dbo.Table WITH (NOLOCK,INDEX=PK_Table) 
SELECT * FROM dbo.Table WITH (INDEX=PK_Table,NOLOCK) 

正則表達式的高級擊穿：

(?=\([^()]*(?=\b(?:NOLOCK|INDEX)\b)[^()]*\)) # a position followed by the words "INDEX" or 
               # "NOLOCK" somewhere in parentheses 
               # (add more keywords if you want) 
(?<!WITH\s+)         # the same position must not be preceded by "WITH" 

這兩個環視條件下準確地確定與關鍵字缺少的位置。之後插入它與-replace運算符很簡單。

詳細擊穿：

(?=      # start zero-width look-ahead 
    \(     # "(" 
    [^()]*     # any character except "(" and ")", repeat 
    \b      # a word boundary 
    (?=     # start non-matching group 
    (?:NOLOCK|INDEX)  #  "NOLOCK" or "INDEX" 
)      # end group 
    \b      # a word boundary 
    [^()]*     # any character except "(" and ")", repeat 
    \)      # ")" 
)      # end look-ahead 
(?<!      # start zero-width negative look-behind 
    WITH     # "WITH" 
    \s+     # any number of whitespace 
)      # end look-begind 

---

話雖這麼說，你的腳本可以更優雅寫成

(Get-Content -Raw D:\script.sql) -replace $expr,"WITH " 

你也可以將輸出重定向到一個新文件

(Get-Content -Raw D:\script.sql) -replace $expr,"WITH " > script_processed.sql 

The Raw parameters ter將文件作爲一個大字符串讀取，而不是按行讀取。在這種情況下，逐行處理將不會有用。