Sed - Multi-Line Pattern matching after finding first part of a pattern

Question

I need to use sed to:

Replace Strings Such as:

tr("Text");

TO

"text";

tr("Textfsafasfsfas safasdfsafas");

TO

"Textfsafasfsfas safasdfsafas";

foo(tr("text")); TO foo("text");

Do Not Replace these Strings:

tr("text").arg(text);

tr("I am some text") .arg(i do stuff);

tr("I am some text") .arg(i do stuff) .arg(I also do stuff);

where it needs to be handled from 1 up to n lines of this.

My current code does not cover the multi-line .arg edge case. Any bright ideas?

Current sed code:

sed -i '/\btr(/{ :a; s/\btr(\([^)]*\))\([^.]\)/\1\2/I; t; N; ba}' $file_t

Edit:

errors << tr("Error: Encountered an invalid token at line: %1, column: %2.") .arg(m_reader.lineNumber()) .arg(m_reader.columnNumber());

is an edge-case that is still an issue at this time.

Answer 1

$ cat ip.txt 
tr("Text");

tr("Textfsafasfsfas
safasdfsafas");

foo(tr("text"));

tr("text").arg(text);

tr("I am some text")
.arg(i do stuff);

tr("I am some text")
.arg(i do stuff)
.arg(I also do stuff);

tr("I am some text")
    .arg(i do stuff);

If file is small enough to hold in memory after slurping all lines as single string,

$ perl -0777 -pe 's/\btr\(([^)]+)\)(?!\s*.arg)/$1/g' ip.txt 
"Text";

"Textfsafasfsfas
safasdfsafas";

foo("text");

tr("text").arg(text);

tr("I am some text")
.arg(i do stuff);

tr("I am some text")
.arg(i do stuff)
.arg(I also do stuff);

tr("I am some text")
    .arg(i do stuff);

If the output is okay, add the -i or -i.bak option for inplace editing

Edit:

Thanks @Costas for pointing out that \n?\s* can be reduced to \s*