X-Git-Url: https://nuage.metani.fr/gitweb/3GPP.git/blobdiff_plain/d788c1f2b3c65a888d59369fdd121ac20da1660b..2c73ed781e61b91e0aef104746e074d19b88e00c:/bin/docx2asn.sh diff --git a/bin/docx2asn.sh b/bin/docx2asn.sh index 373abb9..89bb19f 100755 --- a/bin/docx2asn.sh +++ b/bin/docx2asn.sh @@ -48,12 +48,22 @@ while [ "$#" -gt 0 ]; do continue fi - # On windows, soffice adds CR characters at line ends and produces EN-DASH in Windows-1252 enconding, i.e. 0x96 - # Remove the CR at each line end and convert EN-DASH to normal minus - echo "${1%.docx}.txt" - LC_ALL=C sed -i $'s/\x96/-/g' "${1%.docx}.txt" - LC_ALL=C sed -i $'s/\r$//' "${1%.docx}.txt" + case $platform in + # Replace EN-DASH (UTF-8 encoding) at line start with - + Linux) LC_ALL=C sed -i $'s/^\u2013/-/g' "${1%.docx}.txt" ;; + + # Replace EN-DASH (Windows-1252 encoding) at line start with - + # Remove CR (0x0d) put by libreoffice in windows before each LF (0x0a) + CYGWIN*) LC_ALL=C sed -i $'s/^\x96/-/g; s/\r$//' "${1%.docx}.txt" ;; + *) echo "Unkown platform $platform" + exit 1;; + esac + # Changes things that should not be so in the specification + # Change "-- Cond N3C MP" to "-- Cond N3C-MP" (to respect condition naming rules) + LC_ALL=C sed -i 's/-- Cond N3C MP/-- Cond N3C-MP/' "${1%.docx}.txt" + LC_ALL=C sed -i 's/^N3C MP/N3C-MP/' "${1%.docx}.txt" + LC_ALL=C sed -i 's/-- Cond HOAndServCellAdd,$/-- Cond HOAndServCellAdd/' "${1%.docx}.txt" fi if [[ -e "${1%.docx}.asn" ]] ; then