mirror of
https://github.com/freedomofpress/dangerzone.git
synced 2025-04-28 18:02:38 +02:00
Pass OCR environment variables into container
This commit is contained in:
parent
757a30794d
commit
5e0c521293
2 changed files with 51 additions and 19 deletions
|
@ -167,17 +167,31 @@ class ConvertToPDF(TaskBase):
|
||||||
|
|
||||||
def run(self):
|
def run(self):
|
||||||
self.update_label.emit("Converting pixels to safe PDF")
|
self.update_label.emit("Converting pixels to safe PDF")
|
||||||
args = [
|
|
||||||
"podman",
|
# Build environment variables list
|
||||||
"run",
|
envs = []
|
||||||
"--network",
|
if self.common.settings.get("ocr"):
|
||||||
"none",
|
envs += ["-e", "OCR=1"]
|
||||||
"-v",
|
else:
|
||||||
f"{self.common.pixel_dir.name}:/dangerzone",
|
envs += ["-e", "OCR=0"]
|
||||||
"-v",
|
envs += [
|
||||||
f"{self.common.safe_dir.name}:/safezone",
|
"-e",
|
||||||
"dangerzone",
|
f"OCR_LANGUAGE={self.common.ocr_languages[self.common.settings.get('ocr_language')]}",
|
||||||
"pixels-to-pdf",
|
|
||||||
]
|
]
|
||||||
|
|
||||||
|
args = (
|
||||||
|
[
|
||||||
|
"podman",
|
||||||
|
"run",
|
||||||
|
"--network",
|
||||||
|
"none",
|
||||||
|
"-v",
|
||||||
|
f"{self.common.pixel_dir.name}:/dangerzone",
|
||||||
|
"-v",
|
||||||
|
f"{self.common.safe_dir.name}:/safezone",
|
||||||
|
]
|
||||||
|
+ envs
|
||||||
|
+ ["dangerzone", "pixels-to-pdf",]
|
||||||
|
)
|
||||||
self.execute_podman(args)
|
self.execute_podman(args)
|
||||||
self.task_finished.emit()
|
self.task_finished.emit()
|
||||||
|
|
|
@ -24,16 +24,34 @@ for PAGE in $(seq 1 $NUM_PAGES); do
|
||||||
IMG_WIDTH=$(cat $WIDTH_FILENAME)
|
IMG_WIDTH=$(cat $WIDTH_FILENAME)
|
||||||
IMG_HEIGHT=$(cat $HEIGHT_FILENAME)
|
IMG_HEIGHT=$(cat $HEIGHT_FILENAME)
|
||||||
|
|
||||||
echo "Converting page $PAGE from pixels to searchable PDF"
|
if [ $OCR = "1" ]; then
|
||||||
|
|
||||||
CONVERT_MSGS=$(convert -size "${IMG_WIDTH}x${IMG_HEIGHT}" -depth ${IMG_DEPTH} rgb:"$RGB_FILENAME" png:"$PNG_FILENAME" 2>&1)
|
echo "Converting page $PAGE from pixels to searchable PDF"
|
||||||
if [ $? -ne 0 ]; then
|
|
||||||
die "Page $PAGE conversion failed: $CONVERT_MSGS"
|
CONVERT_MSGS=$(convert -size "${IMG_WIDTH}x${IMG_HEIGHT}" -depth ${IMG_DEPTH} rgb:"$RGB_FILENAME" png:"$PNG_FILENAME" 2>&1)
|
||||||
fi
|
if [ $? -ne 0 ]; then
|
||||||
|
die "Page $PAGE conversion failed: $CONVERT_MSGS"
|
||||||
|
fi
|
||||||
|
|
||||||
|
CONVERT_MSGS=$(tesseract $PNG_FILENAME $OCR_FILENAME pdf -l $OCR_LANGUAGE 2>&1)
|
||||||
|
if [ $? -ne 0 ]; then
|
||||||
|
die "Page $PAGE conversion failed: $CONVERT_MSGS"
|
||||||
|
fi
|
||||||
|
|
||||||
|
else
|
||||||
|
|
||||||
|
echo "Converting page $PAGE from pixels to PDF"
|
||||||
|
|
||||||
|
CONVERT_MSGS=$(convert -size "${IMG_WIDTH}x${IMG_HEIGHT}" -depth ${IMG_DEPTH} rgb:"$RGB_FILENAME" png:"$PNG_FILENAME" 2>&1)
|
||||||
|
if [ $? -ne 0 ]; then
|
||||||
|
die "Page $PAGE conversion failed: $CONVERT_MSGS"
|
||||||
|
fi
|
||||||
|
|
||||||
|
CONVERT_MSGS=$(convert $PNG_FILENAME $PDF_FILENAME 2>&1)
|
||||||
|
if [ $? -ne 0 ]; then
|
||||||
|
die "Page $PAGE conversion failed: $CONVERT_MSGS"
|
||||||
|
fi
|
||||||
|
|
||||||
CONVERT_MSGS=$(tesseract $PNG_FILENAME $OCR_FILENAME pdf 2>&1)
|
|
||||||
if [ $? -ne 0 ]; then
|
|
||||||
die "Page $PAGE conversion failed: $CONVERT_MSGS"
|
|
||||||
fi
|
fi
|
||||||
done
|
done
|
||||||
|
|
||||||
|
|
Loading…
Reference in a new issue