How to get a specific line based on a pattern if found - sed

Question

I am trying to gather the filenames from a very big file depending on if a particular user, in this case windowsdom\nasarchive is found.
I tried running sed -nr "/-{3,}/h; /Path\s*:/H; /windowsdom\\nasarchive\s+Allow\s+FullControl/{x;G;p}" logfilename but it does not bring anything.
-----------------------
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - April 21 - 2.doc
AccessToString : windowsdom\nasarchive Allow FullControl
BUILTIN\Administrators Allow FullControl
NT AUTHORITY\SYSTEM Allow FullControl
BUILTIN\Users Allow ReadAndExecute, Synchronize
-----------------------
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2009\Credit status - Sept. 23 - 59.doc
AccessToString : windowsdom\acl_1 Allow ReadAndExecute, Synchronize
windowsdom\acl_2 Allow Modify, Synchronize
windowsdom\acl_3 Allow ReadAndExecute, Synchronize
windowsdom\adm_server Allow Modify, Synchronize
BUILTIN\Administrators Allow FullControl
-----------------------
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - August 10 - 3.doc
AccessToString : windowsdom\nasarchive Allow FullControl
BUILTIN\Administrators Allow FullControl
NT AUTHORITY\SYSTEM Allow FullControl
BUILTIN\Users Allow ReadAndExecute, Synchronize
-----------------------
Expected result:
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - April 21 - 2.doc
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - August 10 - 3.doc
Can someone think how to get the expected result?

Answer 1

Edit: I know it's not the best idea to edit an accepted answer, but it was substantially inaccurate. It turns out that the hold space is retained between lines.
The main problem with your command is that you are using double quotes, so the escaped backslash is seen unescaped by sed. Change them to single quotes and it starts working:
$ sed -nr '/-{3,}/h; /Path\s*:/H; /windowsdom\\nasarchive\s+Allow\s+FullControl/{x;G;p}' file
-----------------------
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - April 21 - 2.doc
AccessToString : windowsdom\nasarchive Allow FullControl
-----------------------
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - August 10 - 3.doc
AccessToString : windowsdom\nasarchive Allow FullControl
Now, you can simplify it to match the desired output. What you'll eventually get is shown in protong's answer:
sed -rn '/^Path:/h;/windowsdom\\nasarchive\s+Allow\s+FullControl/{g;p}' file
POSIX alternative:
$ sed --posix -n '/^Path:/h;/windowsdom\\nasarchive[[:space:]]\{1,\}Allow[[:space:]]\{1,\}FullControl/{g;p}' log.txt
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - April 21 - 2.doc
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - August 10 - 3.doc

Answer 2

This might work for you (GNU sed):
sed -rn '/^Path:/h;/windowsdom\\nasarchive\s+Allow\s+FullControl/{g;p}' file
This prints the last Path string when it encounters the required string.

Answer 3

Try with awk. You can save each line with the path and print it when found a line which first field matches AccessToString and the third one matches windowsdom\nasarchive:
awk '
$1 ~ /^Path/ { path = $0; next }
$1 ~ /^AccessToString/ && $3 ~ /^windowsdom\\nasarchive$/ { print path }
' infile
It yields;:
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - April 21 - 2.doc
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - August 10 - 3.doc

Answer 4

If you have gawk1 you can define a record as the text between lines of dashes with RS="-----------------------" and each field to be a line by setting FS="\n":
gawk '
BEGIN {RS="-----------------------"; FS="\n"}
$5 ~ /windowsdom\\nasarchive\s+Allow\s+FullControl/ {print $2;}
' ur_file.txt
Prints:
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - April 21 - 2.doc
Path: U:\Credit share\BI-WEEKLY CREDIT NOTES\2010\Credit status - August 10 - 3.doc
1To find out the version of awk, use awk --version. It is typical for linux to have gawk and have that linked to awk. OS X does not have gawk but it is easily installed.