Joining certain lines of text together before modifying them

Question

I have a cricket commentary like -

0.3
Richardson to RG Sharma, 2 runs, gentle, non-swinging length ball on off, Rohit seeks to pick it up. Hit high on the bat and spooned over the returning Mitch Marsh at midwicket
0.4
Richardson to RG Sharma, 1 run, banged in short, controlled pull to long leg

What I want to do is join the n.n lines with the following lines, replacing newlines with spaces. Then, I want to delete everything from the second comma to the end on each line, so the output from the input above should be:

0.3 Richardson to RG Sharma, 2 runs
0.4 Richardson to RG Sharma, 1 run

---

-- Edit -- I used [0-9]"\n"[A-Z] to match the newline but I have trouble reproducing the exact number and character that was matched. Also for deleting everything after 2 commas, I tried parameter expansion but I could only get it to work for deleting after a single comma.

Answer 1

You can use sed:

sed -ne '/^[0-9][0-9]*\.[0-6]/ { N; s/\n/ /; s/^\([^,]*,[^,]*\),.*$/\1/; p; }' < data

This processes a file called data, suppressing printing unless asked for (-n) and executing the sed program in quotes.

That program selects lines starting with one or more digits, a ., and a digit 0-6, and then runs the part in {} for those lines. The N command reads the next line in as well, with a newline in between. The first s/// replaces that newline with a space.

The second matches the new single line up to the second comma and replaces the whole line with just that first part. It captures a \( group \) with some string of non-comma characters, a comma, and some further string of non-comma characters, all followed by a comma and any text .* to the end of the line $.

The final p prints out the resulting line. All other lines of input are ignored.

Answer 2

An alternative solution in AWK:

awk -v FS="," '/^[0-9]/{line=$0;getline; line=line" "$1", "$2 ;print line}' file