Extract text from arbitrary files

convert_to_text.sh

#!/bin/bash

if [ -z "$2" ]; then
  echo "Usage: ./convert_to_text.sh in.pdf out.txt";
  exit 1;
fi

INPUT=$1
OUTPUT=$2
TYPE=`file -b --mime-type "$INPUT"`

if [[ -n `which timeout` ]]; then
  TIMEOUT='timeout -k 4.5 4'
elif [[ -n `which gtimeout` ]]; then
  TIMEOUT='gtimeout -k 4.5 4'
else
  TIMEOUT=''
fi

# file occasionally misidentifies plain text as pascal given the presence of certain keywords :(
if [[ "$TYPE" == "text/plain" || "$TYPE" == "text/x-pascal" ]]; then
  cp "$INPUT" "$OUTPUT";
elif [[ "$TYPE" == "application/pdf" && -n `which pdftotext` ]]; then
  $TIMEOUT pdftotext -q -enc UTF-8 "$INPUT" "$OUTPUT";
elif [[ "$TYPE" == "application/postscript" && -n `which ps2ascii` ]]; then
  $TIMEOUT ps2ascii "$INPUT" "$OUTPUT";
else
  if [[ -n `which abiword` ]]; then
    $TIMEOUT abiword -t txt "$INPUT" -o "$OUTPUT";
  fi
  if [[ ! -f "$OUTPUT" ]] && [[ -n `which unoconv` ]]; then
    $TIMEOUT unoconv -f txt -o "$OUTPUT" "$INPUT";
  fi
fi