CAP Speiseplan Tweetbot

caphro.sh

#!/bin/bash

echo "######################################"
echo "######### CRAWL MENU #################"
echo "######################################"

if [[ $(date '+%u') -eq 1 ]]; then
  DAYFILTER="this"
else
  DAYFILTER="last"
fi

FILTER=$(echo Speiseplan_$(date --date="$DAYFILTER monday" +%d.%m))

REL_PDF_LINK=$(curl "http://www.cap-markt.de/nc/cap-maerkte/cap-marktdetails.html?tx_hgcapmarkt_pi1%5Buser%5D=214&tx_hgcapmarkt_pi1%5Baction%5D=show&tx_hgcapmarkt_pi1%5Bcontroller%5D=User&cHash=b220d37c66fb90e68973a3c05a1fb9df" | grep -o '<a href=['"'"'"][^"'"'"']*pdf['"'"'"]' | sed -e 's/^<a href=["'"'"']//' -e 's/["'"'"']$//' | grep $FILTER)

echo "Download http://www.cap-markt.de$REL_PDF_LINK"

curl "http://www.cap-markt.de$REL_PDF_LINK" > menu.pdf



echo "######################################"
echo "######### CROP AND EXTRACT ###########"
echo "######################################"



echo "precrop"
pdfcrop --margins '-77 -260 -22 -250' menu.pdf precrop.pdf

case "$(date +%u)" in
  "1") pdfcrop --margins '0 0 -398 0' precrop.pdf croped.pdf
  ;;
  "2") pdfcrop --margins '-100 0 -300 0' precrop.pdf croped.pdf
  ;;
  "3") pdfcrop --margins '-200 0 -200 0' precrop.pdf croped.pdf
  ;;
  "4") pdfcrop --margins '-300 0 -100 0' precrop.pdf croped.pdf
  ;;
  "5") pdfcrop --margins '-400 0 0 0' precrop.pdf croped.pdf
  ;;
  *) echo "FAIL"; pdfcrop --margins '0 0 -398 0' precrop.pdf croped.pdf
  ;;
esac

pdftotext croped.pdf
#convert croped.pdf  -resize 1000% -format TIFF  -depth 16 croped.tif
#convert -colorspace rgb -density 400 croped.pdf -resize 400% -monochrome croped.tif
convert -density 300 croped.pdf -quality 100 croped.tif
tesseract -l deu croped.tif output
cat output.txt | tr "\n" " " | sed "s#  #\n#g" | sed '/^\s*$/d' | sed -r '/^.{,7}$/d' > menu_line_by_line_ocr.txt
cat croped.txt | tr "\n" " " | sed "s#  #\n#g" | sed '/^\s*$/d' | sed -r '/^.{,7}$/d' > menu_line_by_line.txt

[ -s menu_line_by_line.txt ] || echo "file is empty" && rm menu_line_by_line.txt && mv menu_line_by_line_ocr.txt menu_line_by_line.txt

echo "######################################"
echo "############ OUTPUT ##################"
echo "######################################"


echo
$EDITOR menu_line_by_line.txt
echo
cat menu_line_by_line.txt
echo
read -p "Menü Okey? y/n" yn
    case $yn in
        [Yy]* ) echo "dann mal los..."; break;;
        [Nn]* ) exit;;
        * ) echo "Please answer yes or no.";;
    esac

cat menu_line_by_line.txt | while read line
  do tweet.sh/tweet.sh tweet "$line"
     sleep 2
  done