CODEWITHSUNDEEP
sedtext-processing
B. Harker
B. Harker

Reputation: 21

Joining specific lines in file

I have a text file (snippet below) containing some public-domain corporate earnings report data, formatted as follows:

Current assets:
Cash and cash equivalents
                                  $ 21,514       $ 21,120
Short-term marketable securities
                                    33,769         20,481
Accounts receivable
                                    12,229         16,849
Inventories
                                     2,281          2,349

and what I'm trying to do (with sed) is the following: if the current line starts with a capital letter, and the next line starts with whitespace, copy the last N characters from the next line into the last N columns of the current line, then delete the next line. I'm doing it this way, because there are other lines in the files that begin with whitespace that I want to ignore. The results should look like the following:

Current assets:
Cash and cash equivalents         $ 21,514       $ 21,120
Short-term marketable securities    33,769         20,481
Accounts receivable                 12,229         16,849
Inventories                          2,281          2,349

The closest I've come to getting what I want is: 

sed -i -r ':a;N;$!ba;s/[^A-Z]*\n([[:space:]])/\1/g' file.txt

and I believe I've got the pattern matching ok, but the subsequent substitution really messes up the alignment of the columns of numbers. When I first started this, this seemed like a simple operation, but hours of searching and experimenting haven't helped. I'm open to any solutions that use something else other than sed, but would prefer to keep it strictly bash. Thank you much!

Upvotes: 2

Views: 49

Answers (1)

potong
potong

Reputation: 58568

This might work for you (GNU sed):

sed -r '/^[[:upper:]]/{N;/\n\s/{h;x;s/\n.*//;s/./ /g;x;G;s/(\n *)(.*)\1$/\2/};P;D}' file

This solution only processes two consecutive lines that start with an upper-case letter and a white space respectively. All other lines are printed as is.

Having gathered the above two lines into the pattern space (PS), a copy is made and stored in the hold space (HS). Processing now swaps to the HS. The second line is removed and the contents of the first turned into spaces. Processing now swaps back to the PS. The HS is appended to the PS and using matching and back references the length of the first line in spaces is subtracted from the combined lines.

The line(s) are printed and then deleted. If the second line did not begin with a space, by use of the P and D commands, it is not deleted but re-appraised by virtue of the regexp at the start of the sed script.

Upvotes: 1

Related Questions