On Sun 19 Jul 2020 at 11:35:31 (+0000), Ajith R wrote: > > > First, there is a somewhat specific question about unspecified > > substitutions. For all I know about these substitutions, you might > > actually need XSLT to do them properly. > > The substitution that I had in mind requires referring to characters based on > their unicode properties like script, block... > > > I think you should absolutely use perl if it makes you happy. > > Unix has a pretty interesting collection of various small tools (which > > "do one thing and do it well" as you may hear), and shells facilitate > > hooking up their outputs and inputs. Almost as if they were made to do > > just that. > > I don't prescribe to using a tool for the sake of happiness. With my limited > knoweldege I want to select one that is adequate to do the job. > The subsitution that I wanted in many text files was deleting text from > languages other than Malayalam,english and punctuation. This required a > program that could match charcters based on their unicode character of block > / script. I didn't find anything to suggest that sed could do that. May be, I > didn't search properly. > Did I miss a utility(including sed) that can do the kind of substitution I > mentioned above?
I've always found exploring new languages very invigorating: the excitement of having new abilities (the modern generation might call them new superpowers), and the pleasure of learning itself. I find it rather a shame that enjoyment isn't at least an important criterion in selecting your programming tools. One of the most pleasurable times in my career was when our Research Computing Advisor thrust a copy of the Green Book into my hands. http://www.math.bas.bg/bantchev/place/snobol/gpp-2ed.pdf After years of Fortran, this was my first experience of a designed computer language, with dynamic allocation of memory, dynamic typing, associative arrays, and its claim to fame, pattern matching. With over 40 years of language development since then, it's probably difficult for anyone now to get quite the same buzz from meeting this language. (BTW I was never a user of SNOBOL4 itself, but the compiled version, SPITBOL, which was much faster, and available to me on IBM OS/360 and its successors.) Its pattern matching was very ahead of its time. I had to move to Perl4 for two reasons: 1) the flight from mainframes in academia, in favour of Unix and PCs, meant that Spitbol became inaccessible for me; 2) Perl, which was developed a few years earlier than Python, had a reasonable DOS/Extended Memory implementation available at the time I got my first PC, a 386DX. Perl gave me an approximation to a Unix shell and its utilities, and the associative arrays that I required, but its pattern matching was much more limited than Spitbol's. By the time Debian was released, I'd been introduced to Python, and decided that it looked more attractive than making the transition Perl4→5. I've never regretted it, despite the hiatuses of the 2→3 and Unicode transitions, particularly as Perl seems to be standing at a fork in the road so much of the time, whereas Python's behaviour appears to have been more decisive. As for your choosing which language to learn, I would spend your time and effort on a language like Perl or Python. You'll probably pick up utilities like grep and sed with the greatest of ease if you're familiar with P⁓. Cheers, David.
