[SOLVED] How can I do a massive find and replace using regular expression?

[fricco]

Hello, I need to find and remove in more than 400 files a specific test. This files are XML files and for example I have a file with the following text:

Code: Select all

    <fieldPermissions>
        <editable>true</editable>
        <field>Quote__c.System_Pin__c</field>
        <readable>true</readable>
    </fieldPermissions>
    <fieldPermissions>
        <editable>true</editable>
        <field>Quote__c.Tax_Type_Client_Fee__c</field>
        <readable>true</readable>
    </fieldPermissions>
    <fieldPermissions>
        <editable>true</editable>
        <field>Quote__c.Tax_Type_Marketing_Fee__c</field>
        <readable>true</readable>
    </fieldPermissions>
    <fieldPermissions>
        <editable>true</editable>
        <field>Quote__c.Tax_Type_Risk_Fee__c</field>
        <readable>true</readable>
    </fieldPermissions>

And remove the complete section that goes:

Code: Select all

    <fieldPermissions>
        <editable>true</editable>
        <field>Quote__c.Tax_Type_Client_Fee__c</field>
        <readable>true</readable>
    </fieldPermissions>

Nowadays I'm using Sublime Text doing a search with the following regular expression:

Code: Select all

\s*<fieldPermissions>\r?\n\s*<editable>(false|true)<\/editable>\r?\n\s*<field>Quote__c.Tax_Type_Client_Fee__c<\/field>\r?\n\s*<readable>(false|true)<\/readable>\r?\n\s*<\/fieldPermissions>\r?\n\s*

This basically freezes Sublime Text as there are lot of files that match. So I'm looking to see if there is a way to do this from command line, expecting to have it done quicker as it doesn't use a UI.
I tried with sed:

Code: Select all

sed "s/\s*<fieldPermissions>\r?\n\s*<editable>(false|true)<\/editable>\r?\n\s*<field>Quote__c.Tax_Type_Client_Fee__c<\/field>\r?\n\s*<readable>(false|true)<\/readable>\r?\n\s*<\/fieldPermissions>\r?\n\s*/something/g"

and perl:

Code: Select all

perl -p -e "s/\s*<fieldPermissions>\r?\n\s*<editable>(false|true)<\/editable>\r?\n\s*<field>Quote__c.Tax_Type_Client_Fee__c<\/field>\r?\n\s*<readable>(false|true)<\/readable>\r?\n\s*<\/fieldPermissions>\r?\n\s*/something/g"

with no luck.
Any clue on how to do this, if it's even possible?

[JerryF]

Check out these posts. I think they may help:
viewtopic.php?f=47&t=240027&p=1314934&h ... e#p1314934

[fricco]

Check out these posts. I think they may help:
viewtopic.php?f=47&t=240027&p=1314934&h ... e#p1314934

Thanks, but in here are talking about UI text editors, which will hav the same issues I'm having with Sublime Text. What I'm looking for is using something from the command line to avoid the load and display of the 400 files

[Lemongrass38]

.

[FreedomTruth]

use sed? http://www.grymoire.com/Unix/Sed.html#uh-51a

Code: Select all

sed '
/<fieldPermissions>/ {
   N
   N
   /Quote__c.Tax_Type_Client_Fee__c/ {
      N
      N
      /<\/fieldPermissions>/d
   }
}
'

You still need to handle input/output but the above code should do what you want, I hope.

[fricco]

use sed? http://www.grymoire.com/Unix/Sed.html#uh-51a

Code: Select all

sed '
/<fieldPermissions>/ {
   N
   N
   /Quote__c.Tax_Type_Client_Fee__c/ {
      N
      N
      /<\/fieldPermissions>/d
   }
}
'

You still need to handle input/output but the above code should do what you want, I hope.

Wow, yes it did! Any idea now how can I use this script to update the same file? When I tried file.xml > file.xml the result was an empty file.xml

[rene]

Use sed -i -e "script" input.xml for in-place editing of input.xml. sed -i.orig -e "script" input.xml for the same while backing up the original input.xml to input.xml.orig.

[WharfRat]

You should be able to incorporate find to go through all of the files e.g., find /usr/share/dbus-1/ -iname *.xml -exec cat '{}' \; replacing cat with your sed directive.

I would copy your target files to another location and test it first.

[lmuserx4849]

Incorporate FreedomTruth's sed into bash.

Code: Select all

#!/bin/bash
declare -- file=''
declare -- dir="$HOME"                              # ** change **
declare -- backup='/tmp/backup/'                # ** change **
declare -- log="/tmp/backup/${0##*/}.log"   # ** change **
declare -- rc=0

if [[ ! -d "${backup}" ]]; then
  mkdir "${backup}"
  printf -- 'Created directory %s\n' "${log}" >> "${log}"
fi

while read -re -d $'\0' file; do

  printf -- 'Processing %s\n' "${file}" >> "${log}"
  cp -a "${file}" "${backup}"

  sed -i '
  /<fieldPermissions>/ {
    N
    N
    /Quote__c.Tax_Type_Client_Fee__c/ {
        N
        N
        /<\/fieldPermissions>/d
    }
  }
  ' "${file}"

  rc=$?
  if (( "${rc}" != 0 )); then
    printf 'sed exit status not 0 [%s]' "${rc}"
    exit 1
  fi

done < <(find "${dir}" -name '*.xml' -print0)   # ** change as needed **

[fricco]

Thanks everyone, great advices. I will try them today and will mark the post as SOLVED once I confirm is working. But again, THANKS!