find a pattern and replace one character of it

Question

Will this be possible in sed ?

I want to replace , in tuple start with [ and end with ] to |

"platform": "iPhone","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"},"screen_flow":["Main","Login"] "platform": "android","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"}

output :

"platform": "iPhone","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"},"screen_flow":["Main"|"Login"] "platform": "android","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"}

If not possible any other solution will be appreciated.

Thank you.

Is the array always in the last value, or can there be several arrays? — Tom Fenech
– Tom Fenech, Commented Aug 20, 2014 at 11:47

Avinash Raj · Accepted Answer · 2014-08-20 11:41:31Z

The below perl code would replace all the , present inside [] braces with | symbol.

$ perl -pe 's/,(?=[^\[\]]*\])/|/g' file "platform": "iPhone","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"},"screen_flow":["Main"|"Login"] "platform": "android","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"}

To save the changes made,

perl -i -pe 's/,(?=[^\[\]]*\])/|/g' file

Jotne · Accepted Answer · 2014-08-20 11:43:21Z

Here is an awk

awk -F"[][]" 'NF>1 {gsub(/,/,"*",$2);$2="["$2"]";sub(/ \[/,"[")}1' file "platform": "iPhone","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"},"screen_flow":["Main"*"Login"] "platform": "android","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"}

NeronLeVelu · Accepted Answer · 2014-08-20 12:07:14Z

sed ':a s/\(\[[^][]*\),\([^][]*]\)/\1|\2/ t a' YourFile

posix version (so --posix with gnu sed) take into accound more than a simple couple between [] and several [] on same line. Need for this e recursif call to change each single occurence one by one until there are no more case.

no back reference like in perl but there are a kind of if/goto that could do the trick with recursive call. Sed have some a logical view of litteral object that is sometime a bite "coming from mars" ;-)

fedorqui · Accepted Answer · 2014-08-20 11:58:30Z

my sed solution:

$ sed 's/[[]\(.\+\),\(.\+\)[]]/[\1|\2]/g' file "platform": "iPhone","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"},"screen_flow":["Main"|"Login"] "platform": "android","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"}

work for sample request, but limited to 1 [] and max 2 field in []. According that, it's fine

clt60 · Accepted Answer · 2014-08-20 12:18:55Z

The next perl:

perl -pE 's/\[(.*?),(.*?)\]/[\1|\2]/g' <<EOF "platform": "iPhone","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"},"screen_flow":["Main","Login"] "platform": "android","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"} EOF

prints:

"platform": "iPhone","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"},"screen_flow":["Main"|"Login"] "platform": "android","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"}

@AvinashRaj, sure ;) it is possible to construct many variations, when it will not works.. E.g. for example, when the last ] missing and so on.. :) or when the comma is missing ... ;)

Community · Accepted Answer · 2017-05-23 11:49:31Z

This is just yet another instance of the issue discussed at https://stackoverflow.com/a/25350541/1745001.

This should do it:

$ awk -F'[][]' -v OFS= 'gsub(/,/,"|",$2){$2="["$2"]"}1' file "platform": "iPhone","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"},"screen_flow":["Main"|"Login"] "platform": "android","report_attrs": {"SKU": "XYZ123","Unit Price": "32.25","Qty": "3"}

Collectives™ on Stack Overflow

find a pattern and replace one character of it

6 Answers 6

Comments

Comments

1 Comment

1 Comment

2 Comments

Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

6 Answers 6

Comments

Comments

1 Comment

1 Comment

2 Comments

Comments

Linked

Related