How to remove line breaks in a numbered list?

Ynglaur · Oct 28, 2014#12014-10-28T22:14+00:00

I converted a PDF into a text file, and the result is that most numbered lists have multiple line breaks. I would like each numbered item to be on its own line, regardless of the line length. I'm trying to insert the lines as records into a database.

Example: I want to convert

Code: Select all

1. some text
 and more text
 and a little more
2. more stuff
3. other stuff
and some
more other stuff

to

Code: Select all

1. some text and more text and a little more
2. more stuff
3. other stuff and some more other stuff

Any help would be greatly appreciated.

Ovg · Oct 29, 2014#22014-10-29T04:19+00:00

Search->Replace
Check Regular expressions: and choose Perl
Check Replace all from top of file

Step One
Find what: \r\n
Replace with: space (symbol with ASCII code 32)

Step Two
Check Regular expressions: and choose Perl
Check Replace all from top of file

Find what: (.)(\d+\.)
Replace with: $1\r\n$2

Mofi · Oct 29, 2014#32014-10-29T06:54+00:00

A perhaps better one step solution. Run a Perl Regular Expression replace searching for [\t ]*\r?\n[\t ]*(?![\r\n\d]) and using a single space as replace string.

Ovg · Oct 30, 2014#42014-10-30T07:11+00:00

Mofi wrote:A perhaps better one step solution.

Yes, indeed!!! Could you please give some brief explanations for "[\t ]*\r?\n[\t ]*(?![\r\n\d])". I nearly understood this regex, but I want full clarity.
Thank you for your help!

Mofi · Oct 30, 2014#52014-10-30T09:06+00:00

[\t ]* ... find zero or more tabs or spaces left to

\r?\n ... an optional carriage return and a line-feed (= DOS or UNIX line terminator, but not MAC) and

[\t ]* ... find zero or more tabs or spaces at beginning of next line

(?![\r\n\d]) ... which starts NOT with a carriage return, line-feed or a digit after optional tabs and spaces. This is a negative lookahead.

This expression does not work for something like

Code: Select all

4. I have a
 6 years old son.
5. His name is John.

The second line starts also with a digit. A lookahead or lookbehind must be of fixed length. Therefore a negative lookahead could not be used to check if there is a dot after the number with 1 or more digits.

But it would be possible to use multiple negative lookaheads. So an even better search string for this task would be [\t ]*\r?\n[\t ]*(?![\r\n])(?!\d\.)(?!\d\d\.)(?!\d\d\d\.) if maximum number of the list (= maximum number of digits, here 3) is known.

Ovg · Oct 30, 2014#62014-10-30T11:41+00:00

Thank you VERY MUCH, Mofi!!! Very useful answer!