後続の行マッチパターンのときに2行を結合する

Question

git log出力をasciidoc形式の変更ログにフォーマットしようとしています。私はgit log --formatを使ってこれを行っています。次に、件名にコミットメッセージのバグ番号を追加する必要があります。私は、これはAwkのを使用して行うことができるかどうかを知りたい

 * This is subject without issue number 
     + 
     There will be multiple lines of text and multiple paragraphs. 

     2nd paragraph of the commit message. 


     * This is commit with issue number 
     + 
     There can be multiple lines of comment message. 

     2nd paragraph of the commit message. A line with Bug: issue ### 
     will be the last line. I need to combine the issue ### with 
     the subject line. 

     Bug: issue 1234 

     * This is commit with issue number in Issue: 1235 format 
     + 
     There can be multiple lines of comment message. 

     2nd paragraph of the commit message. A line with Issue: ### 
     will be the last line. I need to combine the issue ### with 
     the subject line. 

     Issue: 1235 

予想される出力

 * This is subject without issue number 
     + 
     There will be multiple lines of text and multiple paragraphs. 

     2nd paragraph of the commit message. 


     * issue 1234 This is commit with issue number 
     + 
     There can be multiple lines of comment message. 

     2nd paragraph of the commit message. A line with Bug: issue ### 
     will be the last line. I need to combine the issue ### with 
     the subject line. 

     * issue 1235 This is commit with issue number in Issue: 1235 format 
     + 
     There can be multiple lines of comment message. 

     2nd paragraph of the commit message. A line with Issue: ### 
     will be the last line. I need to combine the issue ### with 
     the subject line. 

：以下

入力は

 git log --reverse --no-merges $1..$2 --format='* %s%n+%n%b' | \ 
     sed -e '/^Change-Id:.*$/d' | sed -e '/^Signed-off-by:.*$/d' 

入力例を使用して生成されます。あなたは上記を達成できるAwkコードを提供できますか？他のオプションは何ですか？私は、希望の出力を生成するシェルスクリプトを作成したいと思います。

Answer 1

awk ' 
    $1 == "*" { 
     if (entry) { 
      print subject 
      print entry 
      entry = "" 
     } 
     subject = $0 
     next 
    } 
    $1 == "Bug:" { 
     sub(/\*/, "* issue " $NF, subject) 
     next 
    } 
    {entry = entry "\n" $0} 
    END {print subject; print entry} 
' 

Answer 2

一つの方法は、あなたの入力例は、infileのコンテンツであることが想定SEDの使用（おそらく\sのGNU sedを必要とsedが不平を言うならばリテラルのスペースでそれを変更します。）：

sed -n '/^\s*bug:/I ! { H ; b }; s/^[^:]*:// ; G ; s/\([^\n]*\)\n\(.*\*\)/\2\1/ ; s/^\n// ; p' infile 

出力：

* This is subject without issue number 
+ 
There will be multiple lines of text and multiple paragraphs. 

2nd paragraph of the commit message. 


* issue 1234 This is commit with issue number 
+ 
There can be multiple lines of comment message. 

2nd paragraph of the commit message. A line with Bug: issue ### 
will be the last line. I need to combine the issue ### with 
the subject line. 

説明：

-n       # Disable printing. 
/^\s*bug:/I ! {    # If line doesn't begin with 'bug' ignoring case. 
    H       # Append line to 'hold space'. 
    b       # Read next line. 
} 
s/^[^:]*://     # Line beginning with 'bug': Remove part of line until a colon. 
G       # Get data of 'hold space'. 
s/\([^\n]*\)\n\(.*\*\)/\2\1/ # Put bug line just before the commit message. 
s/^\n//      # Remove leading newline. 
p       # Print. 

Answer 3

ファイル全体がメモリにロードされている場合、それは修正するのは簡単です：

s/^\*((?:(?!^\*).)*)\n^Bug: (issue \d+)\n/* $2$1/msg; 

     ^^^^^^^^^^^^^ 
     Matches anything but "*" at the start of a line. 

そして、これはまさにそれを行います。

perl -0777pe1's/^\*((?:(?!^\*).)*)\n^Bug: (issue \d+)\n/* $2$1/msg'