使用sed命令替换文件中的2行

Question

I try to delete the comma on the next-to-last line:

 { "subject" : { "value" : "http://d-nb.info/gnd/1-2", "type" : "uri" }, "predicate" : { "value" : "http://www.w3.org/1999/02/22-rdf-syntax-ns#type", "type" : "uri" }, "object" : { "value" : "http://d-nb.info/standards/elementset/gnd#SeriesOfConferenceOrEvent", "type" : "uri" } }, { 

And I am using this command:

<test2.file sed '/^\s*\},$/ { N; s/^\s*\},\n\{/^\s*\}\n\{/ }' 

I expect the output should be like that:

 { "subject" : { "value" : "http://d-nb.info/gnd/1-2", "type" : "uri" }, "predicate" : { "value" : "http://www.w3.org/1999/02/22-rdf-syntax-ns#type", "type" : "uri" }, "object" : { "value" : "http://d-nb.info/standards/elementset/gnd#SeriesOfConferenceOrEvent", "type" : "uri" } } { 

But it returns an error.

sed: -e expression #1, char 42: Unmatched {

Does anybody know what's wrong with it? Thx.

PS, Here is the end of the file:

 { "subject" : { "value" : "http://d-nb.info/gnd/1060867974", "type" : "uri" }, "predicate" : { "value" : "http://d-nb.info/standards/elementset/gnd#preferredNameEntityForThePerson", "type" : "uri" }, "object" : { "value" : "_:genid12768016", "type" : "bnode" } } 

Answer 1

don't escape { or } in your pattern/address part. sed uses BRE by default. With BRE, chars like { ( + .. don't have special meaning, you have to escape them to give them special meaning.

In your case, you want to match literal { or } , then don't escape them.

EDIT

I guess you want to remove the last comma, so that change:

...
 },
{

into

...
 }
{

then you can try:

awk -v RS='\0' '7+gsub(/,\s*{/,"\n{")' file

test with your file:

kent$  awk -v RS='\0' '7+gsub(/,\s*{/,"\n{")' f
{
  "subject" : {
    "value" : "http://d-nb.info/gnd/1-2",
    "type" : "uri"
    },
  "predicate" : {
    "value" : "http://www.w3.org/1999/02/22-rdf-syntax-ns#type",
    "type" : "uri"
    },
  "object" : {
    "value" : "http://d-nb.info/standards/elementset/gnd#SeriesOfConferenceOrEvent",
    "type" : "uri"
    }
  }
{

Answer 2

您可以使用缩进识别正确的行吗？

sed 's/^ },/ }/'