Re: anyone have software to OCR a PDF?

[Jerry <SNIPawanderin@yahoSNIPo.ca>]

David Wilson wrote:

> On Feb 13, 9:54 am, David Wilson <mcs6...@gmail.com> wrote:
>> #!/bin/sh
>> tr -c '[:alnum:]' '[\n*62]' | sort | uniq -c | sort -n
>>
>> Hmm - how did those 3 pipes get through the tr?
> 
> The reason is that string2 is too short. In the C locale :alnum: is 62
> characters but the complement of :alnum: is more like 256-62=194
> characters. Replace the 62 with 194 (or even 256) and it works
> properly.

And you can get rid of the "spurious" number (the number of newlines) by
adding an egrep -v to the mix:

tr -c '[:alnum:]' '[\n*194]' | egrep -v '^$' | sort | uniq -c | sort -n

On itself:
      1 194
      1 alnum
      1 bin
      1 egrep
      1 sh
      1 tr
      1 uniq
      1 v
      2 c
      2 n
      2 sort


-- 
--  Jerry