menekis/OCRmyPDF
Public Access
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

OCRmyPDF.sh: fixes #25 and fixes #26

Browse Source 
- In debug mode: compute and echo time required for processing
- Resolutions (x/y) that are nearly equal are not supported (because the
test did not take into account imprecision due to trauncation)
This commit is contained in:
fritz-hh
2013-05-01 15:58:55 +02:00
parent ab07f4deea
commit edaa70b97f
1 changed files with 26 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
OCRmyPDF.sh
+26 -14
View File
@@ -6,6 +6,8 @@
TOOLNAME="OCRmyPDF"
VERSION="v1.0-rc2"
START=`date +%s`
usage() {
cat << EOF
--------------------------------------------------------------------------------------
@@ -150,11 +152,12 @@ cd "`dirname $0`"
today=$(date +"%Y%m%d_%H%M")
fld=$(basename "$FILE_INPUT_PDF" | sed 's/[.][^.]*//')
TMP_FLD="./tmp/$today.filename.$fld"
FILE_TMP="$TMP_FLD/tmp.txt" # temporary file with a very short life used for several things)
FILE_SIZE_PAGES="$TMP_FLD/page-sizes.txt" # size in pt of the respective page of the input PDF file
FILES_OCRed_PDFS="${TMP_FLD}/*-ocred.pdf" # string matching all 1 page PDF files that need to be merged
FILE_OUTPUT_PDF_CAT="${TMP_FLD}/ocred.pdf" # concatenated OCRed PDF files
FILE_OUTPUT_PDFA_WO_META="${TMP_FLD}/ocred-pdfa-wo-metadata.pdf" # PDFA file before appending metadata
FILE_VALIDATION_LOG="${TMP_FLD}/pdf_validation.log" # log file containing the results of the validation of the PDF/A file
FILE_VALIDATION_LOG="${TMP_FLD}/pdf_validation.log" # log file containing the results of the validation of the PDF/A file
# Create tmp folder
[ $VERBOSITY -ge $LOG_DEBUG ] && echo "Creating temporary folder: \"$TMP_FLD\""
@@ -168,22 +171,21 @@ mkdir -p "${TMP_FLD}"
[ $VERBOSITY -ge $LOG_DEBUG ] && echo "Input file: Extracting size of each page (in pt)"
! identify -format "%w %h\n" "$FILE_INPUT_PDF" > "$FILE_SIZE_PAGES" \
&& echo "Could not get size of PDF pages. Exiting..." >&2 && exit $EXIT_BAD_INPUT_FILE
sed -I "" '/^$/d' "$FILE_SIZE_PAGES" # removing empty lines (last one should be)
sed -i "" '/^$/d' "$FILE_SIZE_PAGES" # removing empty lines (last one should be)
numpages=`cat "$FILE_SIZE_PAGES" | wc -l | sed 's/^ *//g'`
[ $VERBOSITY -ge $LOG_INFO ] && echo "Input file: The file has $numpages pages"
numpages=`printf "%04d" $numpages` # add leading zeros to the page number
# Itterate the pages of the input pdf file
cpt="1"
while read pageSize ; do
# add leading zeros to the page number
page=`printf "%04d" $cpt`
[ $VERBOSITY -ge $LOG_INFO ] && echo "Processing page $page"
page=`printf "%04d" $cpt` # add leading zeros to the page number
[ $VERBOSITY -ge $LOG_INFO ] && echo "Processing page $page / $numpages"
# create the name of the required file
curOrigImg="$TMP_FLD/${page}_Image" # original image available in the current PDF page
# (the image file may have a different orientation than in the pdf file)
curHocr="$TMP_FLD/$page.hocr" # hocr file to be generated by the OCR SW for the current page
# (the image file may have a different orientation than in the pdf file)
curHocr="$TMP_FLD/$page.hocr" # hocr file to be generated by the OCR SW for the current page
curOCRedPDF="$TMP_FLD/${page}-ocred.pdf" # PDF file containing the image + the OCRed text for the current page
curOCRedPDFDebug="$TMP_FLD/${page}-debug-ocred.pdf" # PDF file containing data required to find out if OCR worked correctly
@@ -193,7 +195,7 @@ while read pageSize ; do
[ $VERBOSITY -ge $LOG_DEBUG ] && echo "Page $page: size ${heightPDF}x${widthPDF} (h*w in pt)"
# extract raw image from pdf file to compute resolution
# unfortunatelly this image can have another orientation than in the pdf...
# so we will have to extract it again later
# 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`
@@ -215,10 +217,17 @@ while read pageSize ; do
fi
[ $VERBOSITY -ge $LOG_DEBUG ] && echo "Page $page: size ${heightCurImg}x${widthCurImg} (h*w pixel)"
# compute the resolution of the image
dpi_x=$(($widthCurImg*72/$widthPDF))
dpi_y=$(($heightCurImg*72/$heightPDF))
[ "$dpi_x" -ne "$dpi_y" ] && echo "X/Y resolutions not equal ($dpi_x/$dpi_y). Exiting..." >&2 && exit $EXIT_BAD_INPUT_FILE
dpi="$dpi_x"
dpi_x=`echo "scale=5;$widthCurImg*72/$widthPDF" | bc`
dpi_y=`echo "scale=5;$heightCurImg*72/$heightPDF" | bc`
# compute the maximum allowed resolution difference that can be cause by:
# - the truncated PDF with/height in pt
# - the precision of dpi value
epsilon=`echo "scale=5;($widthCurImg*72/$widthPDF^2)+($heightCurImg*72/$heightPDF^2)+0.00002" | bc` # max inaccuracy due to truncation of PDF size in pt
diff1=`echo "($dpi_x - $dpi_y) < $epsilon " | bc`
diff2=`echo "($dpi_y - $dpi_x) < $epsilon " | bc`
[ $diff1 -eq 0 -o $diff2 -eq 0 ] && echo "Resolutions difference ($dpi_x/$dpi_y) higher than expected ($epsilon). Exiting..." >&2 && exit $EXIT_BAD_INPUT_FILE
dpi=`echo "scale=5;($dpi_x+$dpi_y)/2+0.5" | bc` # adding 0.5 is required for rounding
dpi=`echo "scale=0;$dpi/1" | bc` # round to the nearest integer
# Identify if page image should be saved as ppm (color) or pgm (gray)
ext="ppm"
@@ -333,7 +342,7 @@ EOF
[ $VERBOSITY -ge $LOG_DEBUG ] && echo "Output file: Checking compliance to PDF/A standard"
java -jar "$JHOVE" -c "$JHOVE_CFG" -m PDF-hul "$FILE_OUTPUT_PDFA" > "$FILE_VALIDATION_LOG"
grep -i "Status|Message" "$FILE_VALIDATION_LOG" # summary of the validation
[ $VERBOSITY -ge $LOG_DEBUG ] && cat "$FILE_VALIDATION_LOG"
[ $VERBOSITY -ge $LOG_DEBUG ] && echo "The full validation log is available here: \"$FILE_VALIDATION_LOG\""
# check the validation results
pdf_valid=1
grep -i 'ErrorMessage' "$FILE_VALIDATION_LOG" >&2 && pdf_valid=0
@@ -353,5 +362,8 @@ if [ $KEEP_TMP -eq 0 ]; then
fi
END=`date +%s`
[ $VERBOSITY -ge $LOG_DEBUG ] && echo "Script took $(($END-$START)) seconds"
[ $pdf_valid -ne 1 ] && exit $EXIT_INVALID_OUPUT_PDFA || exit 0