sed-删除多个文件中某些字符串的部分

Question

From the tutorial here , I learned to find and replace text across multiple files from the Linux command line. For example,

find . -name "*.php" -print | xargs sed -i 's/foo/bar/g'

finds all instances of 'foo' in files that end with '.php' and replaces such instances with 'bar'.

Now suppose I have multiple, redundant occurrences of something like assert something.makeAssertion that I want to replace with just something.makeAssertion . Given that something can vary, how do I adapt the script for this? I tried some things with $ to capture something as a variable (?) but haven't figured it out.

Answer 1

Superficially, you could use:

find . -name '*.php' -exec \
    sed -i.bak -e 's/assert\(.*\.makeAssertion\)/\1/g' {} +

The regex drops the assert when there's a .makeAssertion later on the line. The find … -exec runs the sed on the files it finds. The + indicates 'group a convenient number of files into a single command line', rather like xargs but without some complications. An alternative using GNU find and xargs would be:

find . -name '*.php' -print0 |
    xargs -0 sed -i.bak -e 's/assert\(.*\.makeAssertion\)/\1/g'

Both these avoid problems with spaces or newlines in file names.