From: David Lecompte <3gpp@metani.eu>
Date: Fri, 5 Jun 2026 07:54:59 +0000 (+0200)
Subject: docx2asn.sh: adjusted removal of EN-dash and CR
X-Git-Url: https://nuage.metani.fr/gitweb/3GPP.git/commitdiff_plain/8440c6232531603286d98ce60510f3ebb00bf55e

docx2asn.sh: adjusted removal of EN-dash and CR
---

diff --git a/bin/docx2asn.sh b/bin/docx2asn.sh
index 373abb9..8a09bb5 100755
--- a/bin/docx2asn.sh
+++ b/bin/docx2asn.sh
@@ -48,12 +48,16 @@ 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	
     fi
 
     if [[ -e "${1%.docx}.asn" ]] ; then