Merge pull request #267 from mediamicroservices/makepdf-cleanup

Makepdf cleanup
mediamicroservices · Sep 4, 2019 · 39df424 · 39df424
2 parents 93f2a7f + 37cea4c
commit 39df424
Show file tree

Hide file tree

Showing 2 changed files with 34 additions and 51 deletions.
diff --git a/makepdf b/makepdf
@@ -1,49 +1,18 @@
 #!/bin/bash
 # makepdf, makes a pdf from an image series
 VERSION="1.0"
-SCRIPTDIR=$(dirname $(which "${0}"))
+SCRIPTDIR="$(dirname "$(which "${0}")")"
 . "${SCRIPTDIR}/mmfunctions" || { echo "Missing '${SCRIPTDIR}/mmfunctions'. Exiting." ; exit 1 ;};
-DEPENDENCIES=(ffmpeg convert pdfjoin tesseract)
+DEPENDENCIES=(ffmpeg pdfjoin tesseract)
 _initialize_make
 
-fix_tif_name(){
-    local tif="${1}"
-    local tifname=$(basename "${TIF}")
-    local tifdir=$(dirname "${TIF}")
-    local basename=$(echo "${tifname}" | cut -d_ -f1 | cut -d. -f2)
-    local pageno=$(echo "${tifname}" | cut -d_ -f2 | cut -d. -f1)
-    local EXTENSION=$(echo "${tifname}" | cut -d_ -f2 | cut -d. -f2)
-    local pageno=$(echo "0000${pageno}" | tail -c 4 | head -c 3)
-    _report -dt "Renaming to fit image sequence naming pattern."
-    local old_tif_name="${TIF}"
-    NEW_TIF_NAME="${tifdir}/${basename}_${pageno}.${EXTENSION}"
-    mv -v -n "${old_tif_name}" "${NEW_TIF_NAME}"
-}
-
-check_tif_name(){
-    local tif="${1}"
-    local tifname=$(basename "${TIF}")
-    local tifdir=$(dirname "${TIF}")
-    local basename=$(echo "${tifname}" | cut -d_ -f2 | cut -d. -f2)
-    local pageno=$(echo "${tifname}" | cut -d_ -f2 | cut -d. -f1)
-    local EXTENSION=$(echo "${tifname}" | cut -d_ -f2 | cut -d. -f1)
-    local pageno=$(echo "0000${pageno}" | tail -c 4 | head -c 3)
-    if [ "$EXTENSION" = "TIF" ] && [ $(echo $pageno | awk '{print length()}') = "3" ] ; then
-        continue
-    else
-        fix_tif_name "${TIF}"
-    fi
-}
-
 _usage(){
     echo
     echo "$(basename "${0}") ${VERSION}"
     echo "This application will create a small pdf file (suitable for sharing online) from a collection of image file or package input with the following options."
     echo "Dependencies: ${DEPENDENCIES[@]}"
-    echo "Usage: $(basename ${0}) [ -d /path/to/deliver/to/ ] fileorpackage1 [ fileorpackage2 ...]"
-    echo "  -d directory ( directory to deliver the resulting file to )"
+    echo "Usage: $(basename "${0}") [ -o /path/to/deliver/to/ ] fileorpackage1 [ fileorpackage2 ...]"
     echo "  -o directory ( directory to write the resulting file to )"
-    echo "  -n (dry-run mode, show the commands that would be run but don't do anything)"
     echo "  -e emailaddress ( send an email about the delivery, only valid if -d is used )"
     echo "  -E emailaddress ( send an email about process outcome )"
     echo "  -h ( display this help )"
@@ -54,16 +23,14 @@ _usage(){
 
 # command-line options to set mediaid and original variables
 OPTIND=1
-while getopts ":o:d:e:E:nh" OPT ; do
+while getopts ":o:e:E:h" OPT ; do
     case "${OPT}" in
         o) OUTPUTDIR_FORCED="${OPTARG}" && _check_outputdir_forced ;;
-        d) DELIVERDIR="${OPTARG}" && _check_deliverdir ;;
         e) EMAILADDRESS_DELIVERY="${OPTARG}" ;;
         E) EMAILADDRESS_OUTCOME="${OPTARG}" ;;
-        n) DRYRUN=true;;
         h) _usage ;;
-        *) echo "bad option -${OPTARG}" ; _usage ;;
         :) echo "Option -${OPTARG} requires an argument" ; _writeerrorlog "makepdf" "The option selected required an argument and none was provided. The script had to exit." ; exit 1 ;;
+        *) echo "bad option -${OPTARG}" ; _usage ;;
     esac
 done
 shift $(( ${OPTIND} - 1 ))
@@ -85,7 +52,7 @@ while [ "${*}" != "" ] ; do
     fi
     OUTPUTDIRTEXT="${INPUT}/objects/access/txt_1"
     _run mkdir -p "${LOGDIR}"
-    exec > >(tee "${LOGDIR}/$(basename "${0}")_$(_get_iso8601_c)_$(basename ${0})_${VERSION}.txt")
+    exec > >(tee "${LOGDIR}/$(basename "${0}")_$(_get_iso8601_c)_$(basename "${0}")_${VERSION}.txt")
     exec 2>&1
 
     #_find_input "${INPUT}"
@@ -108,23 +75,42 @@ while [ "${*}" != "" ] ; do
     TMP_JPG_DIR="${TMP_MAKEPDF_DIR}/jpgs"
     _run mkdir -p "${TMP_MAKEPDF_DIR}" "${TMP_JPG_DIR}" "${OUTPUTDIRTEXT}"
 
-    for TIF in `find "${SOURCEDIR}" -maxdepth 1 -mindepth 1 -iname "*.tif" -type f` ; do
-        check_tif_name "${TIF}"
-        TIF_BASE_NAME="$(basename "${TIF%.*}")"
-        if [ "${NEW_TIF_NAME}" ] ; then
+    for TIF in $(find "${SOURCEDIR}" -maxdepth 1 -mindepth 1 -iname "*.tif" -type f | sort) ; do
+        tifname="$(basename "${TIF}")"
+        _report -dt "Working on ${tifname}..."
+        pageno="$(echo "${tifname}" | cut -d_ -f2 | cut -d. -f1)"
+        # check that tif name adheres to filename pattern and rename if not
+        if [[ "${#pageno}" != "3" ]] ; then
+            tifdir="$(dirname "${TIF}")"
+            basename="$(echo "${tifname}" | cut -d_ -f1 | cut -d. -f2)"
+            EXTENSION="${tifname##*.}"
+            pageno="$(echo "0000${pageno}" | tail -c 4 | head -c 3)"
+            _report -dt "Renaming to fit image sequence naming pattern."
+            NEW_TIF_NAME="${tifdir}/${basename}_${pageno}.${EXTENSION}"
+            mv -v -n "${TIF}" "${NEW_TIF_NAME}"
             TIF="${NEW_TIF_NAME}"
         fi
-        if [[ ! -s "${TMP_JPG_DIR}/$(basename "${TIF%.*}").jpg" ]] ; then
-            ffmpeg -i "${TIF}" -pix_fmt yuvj420p -s 1275x1650 "${TMP_JPG_DIR}/${TIF_BASE_NAME}.jpg"
+        TIF_BASE_NAME="$(basename "${TIF%.*}")"
+        JPG_NAME="${TMP_JPG_DIR}/${TIF_BASE_NAME}.jpg"
+        if [[ ! -s "${JPG_NAME}" ]] ; then
+            ffmpeg -hide_banner -nostdin -i "${TIF}" -pix_fmt yuvj420p -s 1275x1650 "${JPG_NAME}"
         fi
-        tesseract "${TMP_JPG_DIR}/${TIF_BASE_NAME}.jpg" "${TMP_JPG_DIR}/${TIF_BASE_NAME}" -l eng --psm 4 pdf "${SCRIPTDIR}/tesseract.conf"
-        tesseract "${TMP_JPG_DIR}/${TIF_BASE_NAME}.jpg" "${TMP_JPG_DIR}/${TIF_BASE_NAME}" -l eng --psm 4     "${SCRIPTDIR}/tesseract.conf"
+        TESSERACT_CONFIG=(-c tessedit_char_whitelist="abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789!@#$%^\&*(){}[]\|\"':;?/>.<,~\` " -c textord_min_linesize=2.25 -c preserve_interword_spaces=1)
+        tesseract "${JPG_NAME}" "${TMP_JPG_DIR}/${TIF_BASE_NAME}" -l eng --psm 4 "${TESSERACT_CONFIG[@]}" pdf
+        tesseract "${JPG_NAME}" "${TMP_JPG_DIR}/${TIF_BASE_NAME}" -l eng --psm 4 "${TESSERACT_CONFIG[@]}" txt
     done
 
+    _report -dt "Checking for PBCore data"
     SCRIPT_TITLE=$(fmpbcore "${MEDIAID}" | xmlstarlet 'select' -N "p=http://www.pbcore.org/PBCore/PBCoreNamespace.html" -t -v "/p:pbcoreCollection/p:pbcoreDescriptionDocument/p:pbcoreTitle[@titleType='Series']" -o ": " -v "/p:pbcoreCollection/p:pbcoreDescriptionDocument/p:pbcoreTitle[@titleType='Episode']")
+    if [[ -n "${SCRIPT_TITLE}" ]] ; then
+        MIDDLE_OPTIONS+=(--pdftitle "${SCRIPT_TITLE}")
+    fi
     SCRIPT_AUTHOR=$(fmpbcore "${MEDIAID}" | xmlstarlet 'select' -N "p=http://www.pbcore.org/PBCore/PBCoreNamespace.html" -t -m "/p:pbcoreCollection/p:pbcoreDescriptionDocument/p:pbcoreCreator" -v "p:creatorRole" -o ": " -v "p:creator" -o " ; ")
+    if [[ -n "${SCRIPT_AUTHOR}" ]] ; then
+        MIDDLE_OPTIONS+=(--pdfauthor "${SCRIPT_AUTHOR}")
+    fi
 
-    pdfjoin --pdftitle "${SCRIPT_TITLE}" --pdfauthor "${SCRIPT_AUTHOR}" --pdfkeywords "${MEDIAID}" --fitpaper 'false' --rotateoversize 'false' --paper letter $(find "${TMP_JPG_DIR}" -name "*.pdf" | sort | xargs) --outfile "${OUTPUT}"
+    pdfjoin "${MIDDLE_OPTIONS[@]}" --pdfkeywords "${MEDIAID}" --fitpaper 'false' --rotateoversize 'false' --paper letter $(find "${TMP_JPG_DIR}" -name "*.pdf" | sort | xargs) --outfile "${OUTPUT}"
 
     find "${TMP_JPG_DIR}" -name "*.txt" | sort | while read page ; do
         PAGENAME="$(basename "${page%.*}")"

diff --git a/tesseract.conf b/tesseract.conf