haeaeae/ocrmypdf
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Support for -f option

Browse Source 
Fixes #16
This commit is contained in:
fritz-hh
2014-01-04 21:24:33 +01:00
parent 884dbce712
commit 5945454597
2 changed files with 42 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files
OCRmyPDF.sh
+12 -6
View File
@@ -1,6 +1,6 @@
#!/bin/sh
##############################################################################
# Copyright (c) 2013: fritz-hh from Github (https://github.com/fritz-hh)
# Copyright (c) 2013-14: fritz-hh from Github (https://github.com/fritz-hh)
##############################################################################
# Import required scripts
@@ -19,7 +19,7 @@ tesseract engine)
Copyright: fritz from NAS4Free forum
Version: $VERSION
Usage: OCRmyPDF.sh [-h] [-v] [-g] [-k] [-d] [-c] [-i] [-o dpi] [-l language] [-C filename] inputfile outputfile
Usage: OCRmyPDF.sh [-h] [-v] [-g] [-k] [-d] [-c] [-i] [-o dpi] [-f] [-l language] [-C filename] inputfile outputfile
-h : Display this help message
-v : Increase the verbosity (this option can be used more than once) (e.g. -vvv)
@@ -36,6 +36,8 @@ Usage: OCRmyPDF.sh [-h] [-v] [-g] [-k] [-d] [-c] [-i] [-o dpi] [-l language] [-
-o : If the resolution of an image is lower than dpi value provided as argument, provide the OCR engine with
an oversampled image having the latter dpi value. This can improve the OCR results but can lead to a larger output PDF file.
(default: no oversampling performed)
-f : Force to OCR the whole document, even if some page already contain font data
(which should not be the case for PDF files built from scnanned images)
-l : Set the language of the PDF file in order to improve OCR results (default "eng")
Any language supported by tesseract is supported.
-C : Pass an additional configuration file to the tesseract OCR engine.
@@ -43,7 +45,7 @@ Usage: OCRmyPDF.sh [-h] [-v] [-g] [-k] [-d] [-c] [-i] [-o dpi] [-l language] [-
Note: The configuration file must be available in the "tessdata/configs" folder
of your tesseract installation
inputfile : PDF file to be OCRed
outputfile : The PDF/A file to be generated
outputfile : The PDF/A file that will be generated
--------------------------------------------------------------------------------------
EOF
}
@@ -76,10 +78,11 @@ PREPROCESS_CLEAN="0" # 0=no, 1=yes (clean image to improve OCR)
PREPROCESS_CLEANTOPDF="0" # 0=no, 1=yes (put cleaned image in final PDF)
OVERSAMPLING_DPI="0" # do not perform oversampling
PDF_NOIMG="0" # 0=no, 1=yes (generates each PDF page twice, with and without image)
FORCE_OCR="0" # 0=do not force, 1=force (Force to OCR the whole document, even if some page already contain font data)
TESS_CFG_FILES="" # list of additional configuration files to be used by tesseract
# Parse optional command line arguments
while getopts ":hvgkdcio:l:C:" opt; do
while getopts ":hvgkdcio:fl:C:" opt; do
case $opt in
h) usage ; exit 0 ;;
v) VERBOSITY=$(($VERBOSITY+1)) ;;
@@ -89,6 +92,7 @@ while getopts ":hvgkdcio:l:C:" opt; do
c) PREPROCESS_CLEAN="1" ;;
i) PREPROCESS_CLEANTOPDF="1" ;;
o) OVERSAMPLING_DPI="$OPTARG" ;;
f) FORCE_OCR="1" ;;
l) LAN="$OPTARG" ;;
C) TESS_CFG_FILES="$OPTARG $TESS_CFG_FILES" ;;
\?)
@@ -207,11 +211,13 @@ numpages=`tail -n 1 "$FILE_PAGES_INFO" | cut -f1 -d" "`
# OCR each page of the input pdf file
! parallel -q -k --halt-on-error 1 "$OCR_PAGE" "$FILE_INPUT_PDF" "{}" "$numpages" "$TMP_FLD" \
"$VERBOSITY" "$LAN" "$KEEP_TMP" "$PREPROCESS_DESKEW" "$PREPROCESS_CLEAN" "$PREPROCESS_CLEANTOPDF" "$OVERSAMPLING_DPI" "$PDF_NOIMG" "$TESS_CFG_FILES" < "$FILE_PAGES_INFO" \
"$VERBOSITY" "$LAN" "$KEEP_TMP" "$PREPROCESS_DESKEW" "$PREPROCESS_CLEAN" "$PREPROCESS_CLEANTOPDF" "$OVERSAMPLING_DPI" \
"$PDF_NOIMG" "$TESS_CFG_FILES" "$FORCE_OCR" < "$FILE_PAGES_INFO" \
&& exit $?
#while read pageInfo ; do
# ! "$OCR_PAGE" "$FILE_INPUT_PDF" "$pageInfo" "$numpages" "$TMP_FLD" \
# "$VERBOSITY" "$LAN" "$KEEP_TMP" "$PREPROCESS_DESKEW" "$PREPROCESS_CLEAN" "$PREPROCESS_CLEANTOPDF" "$OVERSAMPLING_DPI" "$PDF_NOIMG" "$TESS_CFG_FILES" \
# "$VERBOSITY" "$LAN" "$KEEP_TMP" "$PREPROCESS_DESKEW" "$PREPROCESS_CLEAN" "$PREPROCESS_CLEANTOPDF" "$OVERSAMPLING_DPI" \
# "$PDF_NOIMG" "$TESS_CFG_FILES" "$FORCE_OCR" \
# && exit $?
#done < "$FILE_PAGES_INFO"
src/ocrPage.sh
+30 -18
View File
@@ -2,7 +2,7 @@
##############################################################################
# Script aimed at OCRing a single page of a PDF file
#
# Copyright (c) 2013: fritz-hh from Github (https://github.com/fritz-hh)
# Copyright (c) 2013-14: fritz-hh from Github (https://github.com/fritz-hh)
##############################################################################
. "./src/config.sh"
@@ -22,7 +22,7 @@ PREPROCESS_CLEANTOPDF="${10}" # Put the cleaned paged in the OCRed PDF
OVERSAMPLING_DPI="${11}" # Oversampling resolution in dpi
PDF_NOIMG="${12}" # Request to generate also a PDF page containing only the OCRed text but no image (helpful for debugging)
TESS_CFG_FILES="${13}" # Specific configuration files to be used by Tesseract during OCRing
FORCE_OCR="${14}" # Force to OCR, even if the page already contains fonts
@@ -54,18 +54,24 @@ imageCharacteristics() {
# path of the file in which the output should be written
curImgCharacteristics="$4"
# check if the page already contains fonts (which should not be the case for PDF based on scanned files
[ `pdffonts -f $page -l $page ${FILE_INPUT_PDF} | wc -l` -gt 2 ] && echo "Page $page: This page already contains fonts !!!" && return 1
[ $VERBOSITY -ge $LOG_DEBUG ] && echo "Page $page: Size ${heightPDF}x${widthPDF} (h*w in pt)"
# check if the page already contains fonts (which should not be the case for PDF based on scanned files
[ `pdffonts -f $page -l $page ${FILE_INPUT_PDF} | wc -l` -gt 2 ] && echo "Page $page: Page already contains font data !!!" && return 1
# extract raw image from pdf file to compute resolution
# unfortunately this image can have another orientation than in the pdf...
# so we will have to extract it again later using pdftoppm
pdfimages -f $page -l $page -j "$FILE_INPUT_PDF" "$curOrigImg" 1>&2
# count number of extracted images
nbImg=`ls -1 "$curOrigImg"* | wc -l`
[ $nbImg -ne "1" ] && echo "Page $page: Expecting exactly 1 image on page $page (found $nbImg). Cannot compute dpi value." && return 2
if [ $nbImg -ne "1" ]; then
[ $VERBOSITY -ge $LOG_WARN ] && echo "Page $page: Expecting exactly 1 image on page $page (found $nbImg). Cannot compute dpi value."
return 2
fi
# Get characteristics of the extracted image
curImg=`ls -1 "$curOrigImg"*`
propCurImg=`identify -format "%w %h %[colorspace]" "$curImg"`
@@ -130,13 +136,19 @@ curImgCharacteristics="$TMP_FLD/${page}-img-characteristics.txt" # Detected char
# auto-detect the characteristics of the embedded image
imageCharacteristics "$page" "$widthPDF" "$heightPDF" "$curImgCharacteristics"
#in case the page contains more than one image, warn the user but go on with default parameters
if [ "$?" -eq "1" ]; then
echo "Page $page: No need to OCR this file. Exiting..." && exit $EXIT_BAD_INPUT_FILE
elif [ "$?" -eq "2" ]; then
dpi=300
echo "Page $page: Continuing anyway, assuming a default resolution of $dpi dpi"
ret_code="$?"
# in case the page contains text do not OCR, unless the FORCE_OCR flag is set
if [ "$ret_code" -eq "1" -a "$FORCE_OCR" -eq "0" ]; then
echo "Page $page: Exiting... (Use the -f option to force OCRing, even though fonts are available in the input file)" && exit $EXIT_BAD_INPUT_FILE
elif [ "$ret_code" -eq "1" -a "$FORCE_OCR" -eq "1" ]; then
colorspaceCurImg="sRGB"
dpi=300
[ $VERBOSITY -ge $LOG_WARN ] && echo "Page $page: OCRing anyway, assuming a default resolution of $dpi dpi"
# in case the page contains more than one image, warn the user but go on with default parameters
elif [ "$ret_code" -eq "2" ]; then
colorspaceCurImg="sRGB"
dpi=300
[ $VERBOSITY -ge $LOG_WARN ] && echo "Page $page: Continuing anyway, assuming a default resolution of $dpi dpi"
else
# read the image characteristics from the file
dpi=`cat "$curImgCharacteristics" | cut -f1 -d" "`
@@ -216,12 +228,12 @@ fi
# delete temporary files created for the current page
# to avoid using to much disk space in case of PDF files having many pages
if [ $KEEP_TMP -eq 0 ]; then
rm "$curOrigImg"*.*
rm "$curHocr"
rm "$curImgPixmap"
rm "$curImgPixmapDeskewed"
rm "$curImgPixmapClean"
rm "$curImgCharacteristics"
rm -f "$curOrigImg"*.*
rm -f "$curHocr"
rm -f "$curImgPixmap"
rm -f "$curImgPixmapDeskewed"
rm -f "$curImgPixmapClean"
rm -f "$curImgCharacteristics"
fi
exit 0