feat: moar ocr

This commit is contained in:
Kat Inskip 2025-10-23 20:30:53 -07:00
parent 71383c5b5b
commit e9b291078d
Signed by: kat
GPG key ID: 465E64DECEA8CF0F
3 changed files with 69 additions and 63 deletions

View file

@ -1,65 +1,5 @@
{
pkgs,
lib,
...
}: {
home.packages = let
inherit (lib.meta) getExe;
ocr = pkgs.writeShellScriptBin "ocr" ''
set -euo pipefail
pushd () {
command pushd "$@" > /dev/null
}
popd () {
command popd "$@" > /dev/null
}
args="$(getopt -a -o r: --long rotate: -- "$@")"
ROTATE=""
FORMAT="jpg"
usage(){
cat <<EOF
$0 - OCR helper, uses ocrmypdf which uses Tesseract OCR under the hood!
Usage: $0 <file>
[ -r input | --rotate input ]: Angle to rotate the image by
[ -f input | --format input ]: Intermediary format, defaults to jpg
EOF
exit 1
}
if [ $# -eq 0 ]; then
usage
exit 1
fi
eval set -- "''${args}"
while :
do
case $1 in
-r | --rotate) ROTATE="$2" ; shift 2 ;;
-f | --format) ROTATE="$2" ; shift 2 ;;
--) shift; break ;;
*) >&2 echo Unsupported option: $1
usage ;;
esac
done
INTERMEDIARY="./image.''${FORMAT}"
pushd $(mktemp -d)
if [ -n "''${ROTATE}" ]; then
${getExe pkgs.imagemagick} $1 -rotate "''${ROTATE}" "''${INTERMEDIARY}"
else
${getExe pkgs.imagemagick} $1 "''${INTERMEDIARY}"
fi
${getExe pkgs.ocrmypdf} -q --rotate-pages-threshold 2.0 --clean --output-type=none \
--sidecar tmp.txt --rotate-pages --image-dpi 300 --deskew --output-type pdfa --jobs 4 \
"''${INTERMEDIARY}" - > /dev/null
printf "%s" "$(< tmp.txt)"
popd
'';
in [
ocr
{pkgs, ...}: {
home.packages = [
pkgs.ocr
];
}

13
packages/ocr/default.nix Normal file
View file

@ -0,0 +1,13 @@
{
lib,
writeShellScriptBin,
imagemagick,
ocrmypdf,
}:
writeShellScriptBin "ocr" ''
export PATH="$PATH:${lib.makeBinPath [
imagemagick
ocrmypdf
]}"
exec ${./ocr.sh} "$@"
''

53
packages/ocr/ocr.sh Executable file
View file

@ -0,0 +1,53 @@
#!/usr/bin/env bash
set -euo pipefail
pushd () {
command pushd "$@" > /dev/null
}
popd () {
command popd "$@" > /dev/null
}
args="$(getopt -a -o r:f: --long rotate:,format: -- "$@")"
ROTATE=""
FORMAT="jpg"
usage(){
cat <<EOF
$0 - OCR helper, uses ocrmypdf which uses Tesseract OCR under the hood!
Usage: $0 [ARGS] [FILE]
[ -r input | --rotate input ]: Angle to rotate the image by
[ -f input | --format input ]: Intermediary format, defaults to jpg
EOF
exit 1
}
if [ $# -eq 0 ]; then
usage
fi
eval set -- "${args}"
while :
do
case "$1" in
-r | --rotate) ROTATE="$2" ; shift 2 ;;
-f | --format) FORMAT="$2" ; shift 2 ;;
--) shift; break ;;
*) >&2 echo "Unsupported option: $1"
usage ;;
esac
done
INTERMEDIARY="./image.${FORMAT}"
pushd "$(mktemp -d)"
if [ -n "${ROTATE}" ]; then
magick "$1" -rotate "${ROTATE}" "${INTERMEDIARY}"
else
magick "$1" "${INTERMEDIARY}"
fi
ocrmypdf -q --clean --output-type=none --sidecar tmp.txt \
--image-dpi 300 --deskew --output-type pdfa --jobs 4 \
"${INTERMEDIARY}" - > /dev/null
printf "%s" "$(< tmp.txt)"
popd ""