CODEWITHSUNDEEP
awksed
darknight1906
darknight1906

Reputation: 45

delete lines before and after a string is found using sed

below is my xml file content -

<?xml version="1.0" encoding="UTF-8"?>
<artifactListing>
    <folder id="REPORTMART" path="/Repository Objects" pathAlias="/00"
        modifiedBy="Maria" lastUpdated="1480426973000" description="Hyperion Root Folder"/>
    <folder id="DATASOURCESFOLD"
        path="/Repository Objects/HRInternalFolder/DataSources"
        pathAlias="/00/HRInternal/DataSources" modifiedBy="Maria" lastUpdated="1492814854000"/>
    <folder id="HRINTERNALFOLD"
        path="/Repository Objects/HRInternalFolder"
        pathAlias="/00/HRInternal" modifiedBy="Maria" lastUpdated="1492814854000"/>
    <folder id="00000158e031595b-0000-0782-0ae57730"
        path="/Repository Objects/TRCS" pathAlias="/00/0"
        modifiedBy="demoadmin" lastUpdated="1492814854000" description="TRCS"/>
    <resource id="JavaScriptUpdateResizeOn_dds_js"
        path="/Repository Objects/Administration/Impact Manager/Script Repository"
        pathAlias="/00/Administration/0/Script_Repository"
        modifiedBy="Maria" lastUpdated="1492814880000"
        description="JavaScript Update DDS configuration with Layout Manager"
        name="JavaScriptUpdateResizeOn_dds.js" type="text/im-javascript" size="-1"/>
    <resource id="449cb46e6b4492f3afb8ef693dffb43a90cdd992"
        path="/Security" pathAlias="/02"
        description="Shared Services Administrator"
        name="epm_default_cloud_admin" type="UserPreferences" size="-1"/>
    <resource id="0f62187cf5a8f5aecec7a9879c9e40497d6d8649"
        path="/Security" pathAlias="/02" description="" name="Jacob"
        type="UserPreferences" size="-1"/>
    <resource id="0df02da8548eeef2174c97c2ade67b4c5adc3160"
        path="/Security" pathAlias="/02" description="" name="Henry"
        type="UserPreferences" size="-1"/>
    <resource id="33dca1c0c1c5ae78f67580a76d9c6aba6a172e20"
        path="/Security" pathAlias="/02" description="" name="Susan"
        type="UserPreferences" size="-1"/>
    <resource id="3e182b1ea9376483a38614d916a0b666ef531b6d"
        path="/Security" pathAlias="/02" description="" name="Maria"
        type="UserPreferences" size="-1"/>
    <resource id="0f62187cf5a8f5aecec7a9879c9e40497d6d8649"
        path="/Security" pathAlias="/02" description="" name="Jacques"
        type="UserPreferences" size="-1"/>
    <resource id="0df02da8548eeef2174c97c2ade67b4c5adc3160"
        path="/Security" pathAlias="/02" description="" name="Frank"
        type="UserPreferences" size="-1"/>
    <resource id="PP_3e182b1ea9376483a38614d916a0b666ef531b6d_0"
        path="/Product Preferences" pathAlias="/05"
        description="This is your default Personal Page."
        name="My Personal Page" type="PersonalPageContent" size="-1"/>
</artifactListing>

now using sed I would like to delete entire resource tag if "Susan" string is found within it, other non-Susan resource tag's should not be considered. In this scenario, it's only 1 line before and after string, I've other cases where there are more line within the resource tag.

Upvotes: 1

Views: 129

Answers (2)

RomanPerekhrest
RomanPerekhrest

Reputation: 92854

Using XML parser is the right way for manipulating XML documents.

xmlstarlet solution:

xmlstarlet ed -d '//resource[@name="Susan"]' yourxmlfile

  • ed - edit mode

  • -d - delete action

  • //resource[@name="Susan"] - xpath expression

Upvotes: 1

Weike
Weike

Reputation: 1270

For your scenario, try this:

$ sed -ibak -r '/^\s*<resource/!bend;:loop;N;/\/>.*$/{/Susan/d;bend};bloop;:end' filename

Explains:

  1. /^\s*<resource/!bend: if pattern space does NOT start with ^\s*<resource, jump to the label named end to start a new loop.
  2. :loop: set a label named loop to deal with the whole resource tag.
  3. N: use N command to append a \n and next line into pattern space.
  4. /\/>.*$/{/Susan/d}: if current pattern space ends with />$, which means we have got a complete resource tag in pattern space, then we can deal with it; if this complete resource tag contains Susan, which is your pattern, use d command to delete all contents in pattern space and then jump to label named end to start a new loop.
  5. bloop: use a loop to append the remaining lines of current resource tag into pattern space.
  6. use -ibak to backup the origin file.

P.S: It will ALSO work for other cases where there are more lines within the resource tag.

Upvotes: 0

Related Questions