Grep "Exception" but filter out one specific case, based on previous line

In my application, I have modified all IP addresses, in order not to disturb the actual production system. As a result, my application is throwing lots of exceptions. Those are kept in a logfile, called filename.

I would like to filter the exceptions, but I don't want to see the ones, caused by the modification of the IP addresses.

This sounds very easy, because those exceptions are preceded by a line, containing Failed to connect.

Let's see how to do this:

Filter on exceptions:

grep "Exception" filename

Show also the previous line:

grep -B 1 "Exception" filename

Do not show the lines, containing "Failed to connect":

grep -B 1 "Exception filename | grep -v "Failed to connect"

=> No, this is not what I want: this filters out the lines, containing the words "Failed to connect", but the actual exceptions are still shown. How can I not only filter out the exceptions too?

My filename contents are something like:

... Failed to connect ...
... Exception ...
...
... (lots of these)
...
... <something else than "Failed to connect">
... Exception ...
...
... Failed to connect ...
... Exception ...
...
... (again lots of these)
...

I'm only interested in the lines ... Exception ... who are not preceded by "Failed to connect".

When I press man grep, it ends with:

GNU grep 3.4 ... 2019-12-29

Does anybody have an idea?
Thanks in advance

Using gnu-grep you may be able to do this:

grep -zoP '(?m)^(?!.*Failed to connect).+\R.*Exception.*\R' file

... foo bar baz
... Exception ...

# where file content is
cat file

.. Failed to connect ...
... Exception ...
...
... (lots of these)
...
... foo bar baz
... Exception ...
...
... Failed to connect ...
... Exception ...
...
... (again lots of these)
...

RegEx Demo

Command Details:

• -z: Operate on full file instead of one line at a time
• -o: Return only matched text
• -P: Enable PCRE mode
• (?m): Enable MULTILINE mode
• ^: Match a line start
• (?!.*Failed to connect): Negative lookahead to assert failure when we have Failed to connect anywhere on a line
• .+\R: Match 1+ of any characters followed by a line break
• .*Exception.*\R: Match 0+ of any characters then text Exception then again 0 or more of any characters followed by a line break

With your shown samples, please try following GNU awk code. Written and tested in GNU awk with setting RS to [^\n]*\\n[^E]*Exception[^\n]* and then using match function to only get the required part as per shown output in RT variable of GNU awk.

awk -v RS='[^\n]*\\n[^E]*Exception[^\n]*' '
RT{
  if(RT!~/Failed to connect/ && RT~/Exception/){
    match(RT,/(^|\n)([^\n]*\n[^\n]*$)/,arr)
    sub(/^\n/,"",arr[1])
    print arr[1],arr[2]
  }
}
' Input_file

OR we could make above more efficient as following:

awk -v RS='[^\n]*\\n[^E]*Exception[^\n]*' '
RT{
  if(RT!~/Failed to connect/ && RT~/Exception/){
    match(RT,/(^|\n)([^\n]*\n[^\n]*$)/,arr)
    print arr[2]
  }
}
' Input_file