XPost: uk.comp.os.linux   
   From: java@evij.com.invalid   
      
   On 30/03/2024 16:06, Paul wrote:   
   > On 3/29/2024 6:54 PM, Java Jive wrote:   
   >> As some here may recall, I'm in the midst of several years' work scanning   
   family documents, post-processing them, and then releasing them either to the   
   general public for generations sufficiently removed from living people, or   
   else just back to the    
   source branch of the family. Currently there are nearly 15,000 original   
   documents with equal numbers of post-processed versions arranged as below, but   
   around 7,000 of them completed in previous stages already have final   
   documents, so there are around 8,   
   000 newly added documents left to do.  What I need to do is use a program,   
   feh seems a likely choice, to work through those documents that don't yet have   
   a final version, choose one from five possible versions described below, copy   
   it to its    
   corresponding final directory, and move on to the next.   
   >>   
   >> The documents are arranged as follows:   
   >>   
   >> Family History   
   >>      /_all/1o/subs    Original scans   
   >>      /_all/2n/subs    Normalised versions of the originals   
   >>      /_all/3t/subs    Textcleaned versions ditto   
   >>      /_all/4nt/subs    Normalised then textcleaned versions ditto   
   >>      /_all/5tn/subs    Textcleaned then normalised versions ditto   
   >>      /subs        Final docs chosen from the best of the above   
   >>   
   >> Example for one particular subdirectory:   
   >>   
   >> Family History/_all/1o/Alston   
   >> Family History/_all/2n/Alston   
   >> Family History/_all/3t/Alston   
   >> Family History/_all/4nt/Alston   
   >> Family History/_all/5tn/Alston   
   >> Family History/Alston   
   >>   
   >> I've tried the following command line for feh (sorry can't help any line   
   wrap):   
   >>   
   >> user@computer:~# feh -Fdr -A 'echo cp -a \"%F\" \"$(echo %F | sed -E "{   
   s~/_all/[1-5][ont]{1,2}~~ ; s~ +\([ont]{1,2}\)~~ }" )\"' 'Family   
   History/_all/1o'   
   >>   
   >> ... which results in ...   
   >>   
   >> cp -a "Family History/_all/1o/Alston/Alston Descent - 2nd Earl Of Montrose,   
   William - 1 (o).png" "Family History/Alston/Alston Descent - 2nd Earl Of   
   Montrose, William - 1.png"   
   >>   
   >> ... which looks as though it should do the job, however removing the echo   
   ...   
   >>   
   >> user@computer:~# feh -Fdr -A 'cp -a \"%F\" \"$(echo %F | sed -E "{   
   s~/_all/[1-5][ont]{1,2}~~ ; s~ +\([ont]{1,2}\)~~ }" )\"' 'Family   
   History/_all/1o'   
   >>   
   >> ... results in ...   
   >>   
   >> cp: target '1.png"' is not a directory   
   >>   
   >> Can anyone suggest a magical incantation which will yield the desired   
   result?   
   >>   
   >> Another problem is that I'm going to have order the files so that each of   
   the 5 possible versions are displayed successively so that I can use the arrow   
   keys to go back and forth between them to choose the best.  I think I might   
   be able to use a file    
   list to do this, but as yet I haven't given much thought as to how to create   
   it, it would have to be of the form:   
   >>   
   >> Family History/_all/1o/Alston/Document 1 (o).png   
   >> Family History/_all/2n/Alston/Document 1 (n).png   
   >> Family History/_all/3t/Alston/Document 1 (t).png   
   >> Family History/_all/4nt/Alston/Document 1 (nt).png   
   >> Family History/_all/5tn/Alston/Document 1 (tn).png   
   >> Family History/_all/1o/Alston/Document 2 (o).png   
   >> Family History/_all/2n/Alston/Document 2 (n).png   
   >> Family History/_all/3t/Alston/Document 2 (t).png   
   >> Family History/_all/4nt/Alston/Document 2 (nt).png   
   >> Family History/_all/5tn/Alston/Document 2 (tn).png   
   >> etc.   
   >>   
   >> I could also use Windows if there's a better solution available there, but   
   that seems unlikely to me.   
   >>   
   >> Any help gratefully received.   
   >>   
   >   
   > If cp had a --dry-run, you could obtain what it sees that way.   
   > When it doesn't have such an option, you can debug another way.   
   >   
   > Make yourself a "dumper" to dump the parameters seen on the command line.   
   > You can put "dumper" in place of "cp" and get the parameters that way.   
   >   
   > gcc -o dumper dumper.c   
   >   
   > #include   
   > int main(int argc, char *argv[])   
   > {   
   >      int i;   
   >      printf("Number of arguments = %d\n", argc-1);   
   >      for(i=0; i      {   
   >         if (i == 0) {   
   >             printf("Program name = %s\n", argv[i]);   
   >         } else {   
   >             printf("%s\n",argv[i]);   
   >         }   
   >      }   
   > }   
   >   
   > $ ./dumper   
   > Number of arguments = 0   
   > Program name = dumper   
   >   
   > $ ./dumper a b c   
   > Number of arguments = 3   
   > Program name = dumper   
   > a   
   > b   
   > c   
   >   
   > Your problem will likely involve "escaping" and "how many   
   > levels of shells, subshells, and interpreting are going on".   
   > You know the drill by now :-) It is the "land of the   
   > random number of backslashes" :-)   
   >   
   > Echo is both a built-in for the bash shell, as well   
   > as a /usr/bin/echo existing (maybe some other shell   
   > might need it).   
   >   
   > By inspection, I don't understand why your command   
   > is failing, but shine a flashlight on it and   
   > see if you can figure it out. The usage of "dumper"   
   > should have the same "nested shell count" as using "cp",   
   > that's my hope.   
      
   Thanks Paul, I've copied the above into a folder to keep it.  It's a   
   clever technique which I have used before with Windows BATch files  -  I   
   still have PrintCMD.COM in my Utilities folder!  However, as regards the   
   current problems, as already described in a follow-up post, I solved it   
   another way, by putting the cp command in a one-line shell script.   
      
   In fact, I've now solved all the problems and have my list.  It's 42,940   
   lines along representing at least 9,000 documents, but actually more   
   than that, because photos, maps, sketches, water-colours, etc will not   
   have textcleaned versions, only 'original' and 'normalised' versions.   
      
   Phew!  Better get to it ...   
      
   Here is the FHBlinkList.sh script that created the list:   
      
   #!/bin/sh   
   ME="${0##*/}"   
   DIRY="${0%/*}"   
      
   # The Family History sub-directory   
   FH="~/Family History/_all"   
      
   OLDIFS=${IFS}   
   IFS=$'\n'   
   for A in $(find "${FH}" -name "*.png" | sort)   
   	do   
   		TARGET="$(echo $A | sed -E '{ s~/_all/[1-5][ont]{1,2}~~ ; s~   
   +\([ont]{1,2}\)~~ }' )"   
   		if [ ! -f "${TARGET}" ] || [ "${A}" -nt "${TARGET}" ]   
   			then   
   				echo -e "${TARGET}\t${A}"   
   		fi   
   	done | sort | sed 's/^[^\t]\+\t//' > "${FH}/FHBlinkList.txt"   
   IFS=${OLDIFS}   
      
   --   
      
   Fake news kills!   
      
   I may be contacted via the contact address given on my website:   
   www.macfh.co.uk   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)