#+begin_src sh
dir=/alexi/Documents/mustard/PED1015

#+begin_src python :results output
from pathlib import Path
import subprocess
import fileinput
import os
def mergeFiles(p, fname):
    """ Takes a Path glob expression"""
    allFiles = [x for x in p]
    with open(fname, 'w') as f:
        for line in fileinput.input(files=allFiles):
            f.write(line)
def rmFiles(p):
    for f in p:
        os.remove(f)
p = Path("/alexi/Documents/mustard/PREUX")

for i in $dir/courrier*.pdf
do
    j=${i%.pdf} #bash magic, remove the suffix but keep the directory
    courrier=$dir/courrier$k
    pdftoppm -png $j.pdf $courrier

for f in p.glob("courrier*.pdf"):
    courrier= p / f"courrier{k}"
    subprocess.run(["pdftoppm", "-png", f, courrier])

    for j in $courrier-*.png
    do
        l=${j%.png}
        tesseract -l fra $j $l
    done
    cat $courrier-*.txt > $courrier.txt
    rm $courrier-*.txt
    # Now we remove the header (everyting up to the date)
    # sed -i.bak -n '/^Dijon, le/,$p' $courrier.txt
    # And everything after the footer
    # sed -i.bak '/^BOCAGE SUD.*DIJON CEDEX$/,$d' $courrier.txt
    k=$((k+1))
done
rm $dir/courrier*.png
rm $dir/courrier*.txt.bak

    for png in p.glob(f"courrier{k}*.png"):
         root =png.with_suffix('')
         subprocess.run(["tesseract", "-l", "fra", png, root])
    txt = p.glob(f"courrier{k}-*.txt")
    mergeFiles(txt, p / f"courrier{k}.txt")
    # We have to rewrite the glob....
    rmFiles(p.glob(f"courrier{k}-*.txt"))
    ## Now we remove the header (everyting up to the date)
    ## sed -i.bak -n '/^Dijon, le/,$p' $courrier.txt
    ## And everything after the footer
    ## sed -i.bak '/^BOCAGE SUD.*DIJON CEDEX$/,$d' $courrier.txt
    #k=$((k+1))
rmFiles(p.glob(f"courrier*.png"))
# rm "$dir"/courrier*.txt.bak