如何根据匹配模式用另一个文件中的行替换文本文件中的行？

Question

I want to replace some lines in a file with lines from another file based on matching a pattern. I want to replace every line that begin with "rolOccupant" after finding "# SBD_ING_USER" to the empty line with content from file x

File x

roleOccupant: uid1
roleOccupant: uid2
roleOccupant: uid45
roleOccupant: uid80

Input file

# SDB_ING_USER
objectClass: organizationalRole
objectClass: top
cn: SDB_ING_USER
description: SDB Ing User Role
roleOccupant: uid1
roleOccupant: uid7
roleOccupant: uid67

# SDB_REGISTERY_USER: 
objectClass: organizationalRole
objectClass: top
cn: SDB_REGISTRY_USER
description: SDB Registry Admin Role
roleOccupant: uid2
roleOccupant: uid34
roleOccupant: uid15

OUTPUT file

# SDB_ING_USER
objectClass: organizationalRole
objectClass: top
cn: SDB_ING_USER
description: SDB Ing User Role
roleOccupant: uid1
roleOccupant: uid2
roleOccupant: uid45
roleOccupant: uid80

# SDB_REGISTERY_USER: 
objectClass: organizationalRole
objectClass: top
cn: SDB_REGISTRY_USER
description: SDB Registry Admin Role
roleOccupant: uid2
roleOccupant: uid34
roleOccupant: uid15

Answer 1

awk one-liner

 awk -v RS= -v ORS='\n\n' 'NR==FNR{a=$0;next} /SDB_ING_USER/{sub(/roleOccupant.*/,""); $0=$0 a} 1' fileX file 

-v RS= to set empty line as the Record Separator

FNR==NR{a=a$0; next} FNR==NR{a=a$0; next} : This will store your File X contents in variable a

/# SDB_ING_USER/ {gsub(/roleOccupant.*/,a ORS)} : While iterating over your Input file if record contains # SDB_ING_USER then substitute all lines starting from roleOccupant to end of record with a ORS in other words a "\\n"

Output:

# SDB_ING_USER
objectClass: organizationalRole
objectClass: top
cn: SDB_ING_USER
description: SDB Ing User Role
roleOccupant: uid1
roleOccupant: uid2
roleOccupant: uid45
roleOccupant: uid80

# SDB_REGISTERY_USER:
objectClass: organizationalRole
objectClass: top
cn: SDB_REGISTRY_USER
description: SDB Registry Admin Role
roleOccupant: uid2
roleOccupant: uid34
roleOccupant: uid15

Answer 2

Using awk

One-liner:

awk 'FNR==NR{r=(r!=""?r RS:"")$0;next}/# SDB_ING_USER/{u=1}u && /^roleOccupant:/{next}u && !NF{print r; u=""}1;END{if(u)print r}' file1 file2

Better Readable:

awk '
     FNR==NR{
          r=(r!=""?r RS:"")$0;
          next
     }
     /# SDB_ING_USER/{
          u=1
     }
     u && /^roleOccupant:/{
         next
    }
    u && !NF{
         print r; 
         u=""
    }1
    END{
         if(u)print r
    }
    ' file1 file2

Explanation:

• FNR==NR{r=(r!=""?r RS:"")$0;next} This block we read only file1 and save file1 rows in variable r , separated by record separator RS , FNR==NR will be true, only when awk reads first file.

• /# SDB_ING_USER/{u=1} if line from file2 , contains regexp in /..../ then, set variable u=1

• u && /^roleOccupant:/{next} if variable u is set, and line starts with roleOccupant , skip such line, and go to next record, from file2

• u && !NF{print r; u=""} u && !NF{print r; u=""} if variable u is set, and !NF , meaning empty line ( NF gives no of fields in record, NF=0 means blank line, !0 is 1 , which is true state), then print variable r , and nullify variable u

• }1 1 at the end does default operation, thats print current record/row/line, in current context, this default operation takes only if record not skipped in above, statements.

file1:

$ cat file1
roleOccupant: uid1
roleOccupant: uid2
roleOccupant: uid45
roleOccupant: uid80

file2:

$ cat file2
# SDB_ING_USER
objectClass: organizationalRole
objectClass: top
cn: SDB_ING_USER
description: SDB Ing User Role
roleOccupant: uid1
roleOccupant: uid7
roleOccupant: uid67

# SDB_REGISTERY_USER: 
objectClass: organizationalRole
objectClass: top
cn: SDB_REGISTRY_USER
description: SDB Registry Admin Role
roleOccupant: uid2
roleOccupant: uid34
roleOccupant: uid15

Output:

$ awk 'FNR==NR{r=(r!=""?r RS:"")$0;next}/# SDB_ING_USER/{u=1}u && /^roleOccupant:/{next}u && !NF{print r; u=""}1;END{if(u)print r}' file1 file2
# SDB_ING_USER
objectClass: organizationalRole
objectClass: top
cn: SDB_ING_USER
description: SDB Ing User Role
roleOccupant: uid1
roleOccupant: uid2
roleOccupant: uid45
roleOccupant: uid80

# SDB_REGISTERY_USER: 
objectClass: organizationalRole
objectClass: top
cn: SDB_REGISTRY_USER
description: SDB Registry Admin Role
roleOccupant: uid2
roleOccupant: uid34
roleOccupant: uid15

Answer 3

Where the input file is sdb.txt and file x is x.txt , I use a sed command file cmd.sed to do this:

sed -f cmd.sed sdb.txt

sed.cmd :

# Within the section from SDB_ING_USER to a blank line
/SDB_ING_USER/,/^$/{
    # Delete roleOccupant lines
    /roleOccupant/d
    # At the blank line
    /^$/{
        # Read x.txt into the output stream
        r x.txt
        # Append a blank line to the output stream
        a
        # Delete the blank line from the input stream
        # (Prevents a blank line before the x.txt content)
        d
    }
}

This is done with a sed command file rather than as a single line sed command bcause the sed r and a commands require a newline, so they are painful to try to use in a one-liner. (See EDIT below regarding newlines and -e )

Output:

$ sed -f cmd.sed sdb.txt
# SDB_ING_USER
objectClass: organizationalRole
objectClass: top
cn: SDB_ING_USER
description: SDB Ing User Role
roleOccupant: uid1
roleOccupant: uid2
roleOccupant: uid45
roleOccupant: uid80

# SDB_REGISTERY_USER: 
objectClass: organizationalRole
objectClass: top
cn: SDB_REGISTRY_USER
description: SDB Registry Admin Role
roleOccupant: uid2
roleOccupant: uid34
roleOccupant: uid15

EDIT. Looked at potong's use of e cat which allows reading the x.txt file into the pattern space (rather than output stream) AND realized that at the command line you can just use multiple -e arguments to get effective line breaks in one-liner sed commands.

So here's a new one-liner:

sed -e '/SDB_ING_USER/,/^$/{' -e '/roleOccupant/d' -e '/^$/e cat x.txt' -e '}' sdb.txt

Answer 4

This might work for you (GNU sed):

sed '/SDB_ING_USER/,/^\s*$/!b;/roleOccupant/d;/^\s*$/e cat x' file

Focus on the lines between SDB_ING_USER and the next empty line. Delete any lines containing the string roleOccupant and insert file x before the empty line.